[ { "id": "article:agent-auditable-research", "type": "article", "label": "The Future of Publishing Is Agent-Auditable Research", "path": "/articles/agent-auditable-research/", "maturity": "seed" }, { "id": "article:agent-auditable-research:claim-001", "type": "claim", "label": "Polished prose is becoming cheap; inspectable reasoning is becoming scarce.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:agent-auditable-research:claim-002", "type": "claim", "label": "A future-ready publishing artifact should pair a human essay with a claim graph, evidence ledger, provenance, revision history, and agent contribution record.", "localId": "claim-002", "confidence": "medium-high", "status": "proposal" }, { "id": "article:agent-auditable-research:claim-003", "type": "claim", "label": "Research is the best first wedge because the audience already values citations, methods, uncertainty, and credibility.", "localId": "claim-003", "confidence": "medium", "status": "argument" }, { "id": "article:agent-auditable-research:claim-004", "type": "claim", "label": "Existing publishing, AI research, and protocol tools solve important pieces but not the combined readable-plus-auditable source object.", "localId": "claim-004", "confidence": "medium-high", "status": "landscape" }, { "id": "article:agent-auditable-research:claim-005", "type": "claim", "label": "AI assistance should be disclosed while the human remains accountable for thesis, source selection, wording, and conclusions.", "localId": "claim-005", "confidence": "high", "status": "normative" }, { "id": "article:agent-auditable-research:claim-006", "type": "claim", "label": "Attention-aware reading and machine-readable structure are compatible when the page uses progressive disclosure.", "localId": "claim-006", "confidence": "medium-high", "status": "design" }, { "id": "article:agent-control-planes", "type": "article", "label": "From Agent Swarms to Agent Control Planes", "path": "/articles/agent-control-planes/", "maturity": "contested" }, { "id": "article:agent-control-planes:claim-001", "type": "claim", "label": "Agent orchestration is shifting from hand-written workflows toward governed control planes that route across models, tools, memory, evaluators, policies, and execution environments.", "localId": "claim-001", "confidence": "medium", "status": "core" }, { "id": "article:agent-control-planes:claim-002", "type": "claim", "label": "Mixture-of-Experts and conditional computation predate LLMs and provide the earliest architectural precedent for learned routing.", "localId": "claim-002", "confidence": "high", "status": "core" }, { "id": "article:agent-control-planes:claim-003", "type": "claim", "label": "Learned routers such as FrugalGPT and RouteLLM can match or exceed single-model accuracy at a fraction of the cost, though benchmark caveats apply.", "localId": "claim-003", "confidence": "medium", "status": "core" }, { "id": "article:agent-control-planes:claim-004", "type": "claim", "label": "Test-time search strategies—self-consistency, Tree of Thoughts, Reflexion, and multi-agent debate—expand what a control plane can spend compute on at runtime.", "localId": "claim-004", "confidence": "medium", "status": "argument" }, { "id": "article:agent-control-planes:claim-005", "type": "claim", "label": "Recent learned orchestrators such as Sakana Fugu, Trinity, and Conductor are signals of automated scaffold generation, not settled production recipes.", "localId": "claim-005", "confidence": "medium", "status": "argument" }, { "id": "article:agent-control-planes:claim-006", "type": "claim", "label": "Production control planes combine routing, fallback, policy, memory, evaluation, observability, and lifecycle governance, but no single vendor owns all of them.", "localId": "claim-006", "confidence": "medium", "status": "argument" }, { "id": "article:agent-control-planes:claim-007", "type": "claim", "label": "The term 'control plane' is contested across vendors; teams should judge products by concrete capabilities rather than marketing labels.", "localId": "claim-007", "confidence": "medium", "status": "argument" }, { "id": "article:agent-control-planes:claim-008", "type": "claim", "label": "Teams should treat model selection, fallback, observability, and policy enforcement as infrastructure concerns rather than per-agent code.", "localId": "claim-008", "confidence": "medium", "status": "argument" }, { "id": "article:agentic-commerce-product-truth", "type": "article", "label": "Agentic Commerce and the Product Truth Layer", "path": "/articles/agentic-commerce-product-truth/", "maturity": "seed" }, { "id": "article:agentic-commerce-product-truth:claim-001", "type": "claim", "label": "Modern online commerce is still largely organized around human attention, even when AI is used behind the scenes for targeting, ranking, and recommendation.", "localId": "claim-001", "confidence": "medium-high", "status": "landscape" }, { "id": "article:agentic-commerce-product-truth:claim-002", "type": "claim", "label": "Some brand loyalty is actually status quo bias plus choice overload: the customer sticks with a known product because the market has made exploration expensive.", "localId": "claim-002", "confidence": "medium-high", "status": "behavioral" }, { "id": "article:agentic-commerce-product-truth:claim-003", "type": "claim", "label": "One plausible next commerce shift is from the attention economy to delegated-intent commerce: agents may increasingly translate user preferences into product discovery and purchase decisions.", "localId": "claim-003", "confidence": "medium", "status": "forecast" }, { "id": "article:agentic-commerce-product-truth:claim-004", "type": "claim", "label": "Agentic commerce is likely to need a product assurance layer richer than current product structured data, because agents need evidence, constraints, provenance, and user-fit signals rather than only titles, offers, ratings, and images.", "localId": "claim-004", "confidence": "medium", "status": "proposal" }, { "id": "article:agentic-commerce-product-truth:claim-005", "type": "claim", "label": "The healthiest version of agentic commerce is an open product-truth commons: a contestable, forkable, provenance-rich vocabulary for product claims, evidence, reviews, and buyer-agent preferences.", "localId": "claim-005", "confidence": "medium", "status": "normative" }, { "id": "article:agentic-commerce-product-truth:claim-006", "type": "claim", "label": "Post-purchase feedback can evolve from star ratings into structured experience packets that preserve human judgment while making outcomes legible to agents.", "localId": "claim-006", "confidence": "medium", "status": "proposal" }, { "id": "article:agentic-commerce-product-truth:claim-007", "type": "claim", "label": "B2B agentic buying may move slower across complex purchases, but narrow recurring procurement categories can be stronger MVP wedges because outcomes are measurable.", "localId": "claim-007", "confidence": "medium-high", "status": "forecast" }, { "id": "article:agentic-commerce-product-truth:claim-008", "type": "claim", "label": "The strategic opportunity is not necessarily to replace commerce incumbents, but to demonstrate a concrete agentic behavior, such as buyer-aligned product switching backed by evidence packets and post-purchase feedback, that incumbents may later adopt, adapt, or standardize around.", "localId": "claim-008", "confidence": "medium", "status": "strategy" }, { "id": "article:agentic-commerce-product-truth:claim-009", "type": "claim", "label": "Agentic product trust should shift from review aggregation to adversarial claim ledgers: each product claim should carry source, scope, evidence, counter-evidence, incentive, expiry, and dispute state.", "localId": "claim-009", "confidence": "medium", "status": "proposal" }, { "id": "article:agentic-commerce-product-truth:claim-010", "type": "claim", "label": "A fair product-truth commons needs graded evidence tiers so offline buyers and small sellers can participate without pretending every attestation has the same trust weight.", "localId": "claim-010", "confidence": "medium", "status": "proposal" }, { "id": "article:agentic-commerce-product-truth:claim-011", "type": "claim", "label": "Product-truth infrastructure can reduce the value of fake reviews, but it cannot eradicate manipulation; it moves the battlefield from cheap text generation to collusion, credential abuse, data access, privacy leakage, and governance capture.", "localId": "claim-011", "confidence": "medium-high", "status": "risk" }, { "id": "article:agentic-commerce-product-truth:claim-012", "type": "claim", "label": "Private review entitlements should separate purchase or use verification from public identity: the public system should verify an unlinkable, one-time entitlement token rather than linking a review to a user, receipt, store, or account.", "localId": "claim-012", "confidence": "medium", "status": "proposal" }, { "id": "article:agentic-commerce-product-truth:claim-013", "type": "claim", "label": "Agentic commerce should expose a dual evidence surface: machine-readable claim ledgers for agents and human-readable media, social, community, and brand context for final human judgment.", "localId": "claim-013", "confidence": "medium", "status": "proposal" }, { "id": "article:agentic-commerce-product-truth:claim-014", "type": "claim", "label": "Agentic product assurance should be built around signed, scoped, contestable claims about specific product identities, not aggregate reviews or universal truth labels.", "localId": "claim-014", "confidence": "medium-high", "status": "framing" }, { "id": "article:agentic-commerce-product-truth:claim-015", "type": "claim", "label": "Robust agentic product assurance needs infrastructure beyond reviews and credentials: product identity/versioning, recall feeds, liability, auditors, decision receipts, dispute propagation, portability, red-team benchmarks, and accessible presentation.", "localId": "claim-015", "confidence": "medium", "status": "proposal" }, { "id": "article:agents", "type": "article", "label": "Agents: Goal-Directed AI Systems That Use Tools", "path": "/articles/agents/", "maturity": "seed" }, { "id": "article:agents:claim-001", "type": "claim", "label": "An AI agent is a system that pursues a goal across multiple steps, choosing when to use tools, what to remember, and when to stop.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:agents:claim-002", "type": "claim", "label": "The idea of an agent that follows goals and uses tools is older than large language models; it appears in automation scripts, personal assistants, and game AI.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:agents:claim-003", "type": "claim", "label": "In practice, an agent's loop repeatedly decides which tool to use, what to remember, and whether the goal is satisfied.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:agents:claim-004", "type": "claim", "label": "Agent behavior depends heavily on clear goals, reliable tools, and careful limits; without them, autonomy becomes cost and error.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:agents:claim-005", "type": "claim", "label": "A modern AI agent can be understood as a model plus a harness that provides tools, memory, permissions, checkpoints, and human oversight.", "localId": "claim-005", "confidence": "medium-high", "status": "design" }, { "id": "article:ai-agent-advanced-questions", "type": "article", "label": "Beyond the First Conversation: Advanced Questions for New AI Agent Users", "path": "/articles/ai-agent-advanced-questions/", "maturity": "seed" }, { "id": "article:ai-agent-advanced-questions:claim-001", "type": "claim", "label": "A short concrete privacy checklist is usually more practical for new AI users than a long explanation of how training data works.", "localId": "claim-001", "confidence": "medium-high", "status": "core" }, { "id": "article:ai-agent-advanced-questions:claim-002", "type": "claim", "label": "Teaching new users three recovery moves — ask for sources, rephrase, and test with a known answer — is enough to turn a wrong answer from a stop sign into a learning moment.", "localId": "claim-002", "confidence": "medium", "status": "argument" }, { "id": "article:ai-agent-advanced-questions:claim-003", "type": "claim", "label": "A simple low-stakes versus high-stakes framing is enough to help non-technical users decide when to verify AI output.", "localId": "claim-003", "confidence": "medium-high", "status": "core" }, { "id": "article:ai-agent-advanced-questions:claim-004", "type": "claim", "label": "Asking the agent for a brief summary at the end of a session is the easiest way for a beginner to preserve context across multiple conversations.", "localId": "claim-004", "confidence": "medium", "status": "design" }, { "id": "article:ai-agent-advanced-questions:claim-005", "type": "claim", "label": "Four plain-language moves — context, desired output, exclusions, and options — are enough to improve most beginner prompts without teaching prompt-engineering jargon.", "localId": "claim-005", "confidence": "medium-high", "status": "core" }, { "id": "article:ai-agent-advanced-questions:claim-006", "type": "claim", "label": "A repeated \"doing\" prompt saved as a reusable template is the simplest form of automation for non-technical AI users.", "localId": "claim-006", "confidence": "medium", "status": "argument" }, { "id": "article:ai-agent-advanced-questions:claim-007", "type": "claim", "label": "A short task-based comparison table is more useful to non-technical readers than benchmark scores or feature lists.", "localId": "claim-007", "confidence": "medium", "status": "design" }, { "id": "article:ai-agent-first-conversation", "type": "article", "label": "You Do Not Need to Learn AI First: A 5-Minute Conversation Recipe", "path": "/articles/ai-agent-first-conversation/", "maturity": "seed" }, { "id": "article:ai-agent-first-conversation:claim-001", "type": "claim", "label": "A single model-agnostic starter prompt is a more effective onboarding artifact for non-technical users than a feature list or vendor tutorial.", "localId": "claim-001", "confidence": "medium-high", "status": "core" }, { "id": "article:ai-agent-first-conversation:claim-002", "type": "claim", "label": "The AI agent itself can act as the tutor, so newcomers do not need to study AI before they start using it.", "localId": "claim-002", "confidence": "medium-high", "status": "core" }, { "id": "article:ai-agent-first-conversation:claim-003", "type": "claim", "label": "Suggesting the next question or direction after each answer removes the blank-page problem and keeps the experience experiential rather than instructional.", "localId": "claim-003", "confidence": "medium", "status": "argument" }, { "id": "article:ai-agent-first-conversation:claim-004", "type": "claim", "label": "Two short annotated transcripts are enough to teach the pattern: one proving intelligence through document explanation, and one proving productivity through one-input-multiple-outputs automation.", "localId": "claim-004", "confidence": "medium", "status": "design" }, { "id": "article:ai-delegation-orchestration", "type": "article", "label": "AI Delegation Orchestration: A Series on Durable Agent Work", "path": "/articles/ai-delegation-orchestration/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-01-from-conversation-to-delegation", "type": "article", "label": "From Conversation to Delegation: Why AI Work Needs a Durable Record", "path": "/articles/ai-delegation-orchestration-01-from-conversation-to-delegation/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-01-from-conversation-to-delegation:claim-001", "type": "claim", "label": "Conversation can remain the interface for AI, but consequential AI work should be organized around bounded delegations rather than chat transcripts.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration-02-the-delegation-record", "type": "article", "label": "The Delegation Record: A Schema for Consequential AI Work", "path": "/articles/ai-delegation-orchestration-02-the-delegation-record/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-02-the-delegation-record:claim-001", "type": "claim", "label": "A delegation record is the system-of-record artifact that makes AI work inspectable, resumable, reviewable, and bounded.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration-03-the-operator-cockpit-problem", "type": "article", "label": "The Operator Cockpit Problem: Why More Traces Are Not Enough", "path": "/articles/ai-delegation-orchestration-03-the-operator-cockpit-problem/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-03-the-operator-cockpit-problem:claim-001", "type": "claim", "label": "The operator problem is not lack of information; it is lack of control routing across many active delegations.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration-04-control-loci-not-human-managers", "type": "article", "label": "Control Loci, Not Human Managers: An Agent-Native Routing Model", "path": "/articles/ai-delegation-orchestration-04-control-loci-not-human-managers/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-04-control-loci-not-human-managers:claim-001", "type": "claim", "label": "Agent systems should route decisions to the right control locus instead of copying human management structures or escalating every uncertainty to a person.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration-05-long-running-delegations", "type": "article", "label": "Long-Running Delegations: How Agents Can Work for Hours Without Losing the Plot", "path": "/articles/ai-delegation-orchestration-05-long-running-delegations/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-05-long-running-delegations:claim-001", "type": "claim", "label": "Long-running AI work is viable only when the delegation has checkpoints, evidence gates, self-remediation loops, interruption rules, rollback paths, and explicit stop conditions.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration-06-capability-contracts-for-agent-networks", "type": "article", "label": "Capability Contracts for Agent Networks", "path": "/articles/ai-delegation-orchestration-06-capability-contracts-for-agent-networks/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-06-capability-contracts-for-agent-networks:claim-001", "type": "claim", "label": "Agent systems should be organized around replaceable capabilities with explicit contracts, not around vague role prompts or artificial replicas of human organizations.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration-07-commitment-boundaries-in-high-stakes-domains", "type": "article", "label": "Commitment Boundaries in High-Stakes Domains", "path": "/articles/ai-delegation-orchestration-07-commitment-boundaries-in-high-stakes-domains/", "maturity": "sprout" }, { "id": "article:ai-delegation-orchestration-07-commitment-boundaries-in-high-stakes-domains:claim-001", "type": "claim", "label": "High-stakes AI use should be designed around commitment boundaries: AI may help prepare work, but external, legal, financial, public, or rights-affecting actions need stricter evidence, review, appeal, privacy, and accountability controls.", "localId": "claim-001", "confidence": "medium", "status": "argument" }, { "id": "article:ai-delegation-orchestration:claim-001", "type": "claim", "label": "AI interfaces can stay conversational, but consequential AI work should be governed through durable delegations, explicit records, operator control surfaces, and domain-specific review boundaries rather than chat transcripts alone.", "localId": "claim-001", "confidence": "medium", "status": "framing" }, { "id": "article:ai-demystified", "type": "article", "label": "AI, De-Mystified: A Field Guide to Modern AI Terminology", "path": "/articles/ai-demystified/", "maturity": "seed" }, { "id": "article:ai-demystified:claim-001", "type": "claim", "label": "AI terminology becomes more useful when each term is explained through plain language, older related ideas, practical examples, benefits, limits, and academic roots instead of being treated as a fresh breakthrough every time.", "localId": "claim-001", "confidence": "medium-high", "status": "core" }, { "id": "article:ai-demystified:claim-002", "type": "claim", "label": "A repeating article structure makes it easier for readers to learn one concept at a time and compare new terms with ideas they already know.", "localId": "claim-002", "confidence": "medium", "status": "design" }, { "id": "article:ai-demystified:claim-003", "type": "claim", "label": "Keeping the series inside Aura Knowledge, focused on one concept at a time, protects it from becoming a broad encyclopedia, benchmark hub, or product marketing guide.", "localId": "claim-003", "confidence": "medium", "status": "strategy" }, { "id": "article:ai-winters-expert-systems", "type": "article", "label": "Winters, Expert Systems, and the Cost of Overpromising Intelligence", "path": "/articles/ai-winters-expert-systems/", "maturity": "seed" }, { "id": "article:ai-winters-expert-systems:claim-001", "type": "claim", "label": "Public evaluation reports such as ALPAC and Lighthill mattered because they tested AI-adjacent promises against measurable usefulness, not because they proved intelligence research was worthless.", "localId": "claim-001", "confidence": "medium-high", "status": "argument" }, { "id": "article:ai-winters-expert-systems:claim-002", "type": "claim", "label": "The phrase 'AI winter' should be handled as a contested historical label for reduced confidence, funding, and commercial enthusiasm, not as proof that AI research stopped.", "localId": "claim-002", "confidence": "medium", "status": "framing" }, { "id": "article:ai-winters-expert-systems:claim-003", "type": "claim", "label": "Expert systems produced useful results in narrow domains where domain knowledge could be encoded and maintained.", "localId": "claim-003", "confidence": "high", "status": "core" }, { "id": "article:ai-winters-expert-systems:claim-004", "type": "claim", "label": "Expert-system limits included knowledge acquisition, updating, evaluation, user trust, and workflow integration, not only inference algorithms.", "localId": "claim-004", "confidence": "medium-high", "status": "landscape" }, { "id": "article:ai-winters-expert-systems:claim-005", "type": "claim", "label": "The durable lesson for modern AI is that intelligence claims need grounded tests, maintenance plans, and institution-aware deployment criteria.", "localId": "claim-005", "confidence": "medium", "status": "argument" }, { "id": "article:audio-first-voice-consumption", "type": "article", "label": "Listening to the Firehose: Can Voice-First, Two-Way Audio Become a Legitimate Assistive Medium?", "path": "/articles/audio-first-voice-consumption/", "maturity": "sprout" }, { "id": "article:audio-first-voice-consumption:claim-001", "type": "claim", "label": "Heavy screen use is widespread among working-age adults and is associated with significant productivity and wellbeing costs, including digital eye strain.", "localId": "claim-001", "confidence": "medium", "status": "core" }, { "id": "article:audio-first-voice-consumption:claim-002", "type": "claim", "label": "Listening and reading impose different cognitive demands; audio is generally more transient and pace-dependent, making it a complement rather than a drop-in replacement for reading.", "localId": "claim-002", "confidence": "medium", "status": "core" }, { "id": "article:audio-first-voice-consumption:claim-003", "type": "claim", "label": "End-to-end, full-duplex spoken dialogue models have moved from research demos to publicly documented systems with low enough latency for natural turn-taking.", "localId": "claim-003", "confidence": "medium", "status": "core" }, { "id": "article:audio-first-voice-consumption:claim-004", "type": "claim", "label": "The value of voice-first audio agents depends more on interaction design—turn-taking, interruption, proactivity, and user agency—than on raw conversational naturalness.", "localId": "claim-004", "confidence": "medium", "status": "argument" }, { "id": "article:audio-first-voice-consumption:claim-005", "type": "claim", "label": "Proactive and always-listening audio agents risk intrusiveness and attention capture; user-initiated or notification-triggered sessions better preserve agency.", "localId": "claim-005", "confidence": "medium", "status": "argument" }, { "id": "article:audio-first-voice-consumption:claim-006", "type": "claim", "label": "Voice-first curation can amplify filter-bubble dynamics in a channel with fewer visual cues for verification, so transparency and user control over selection are essential.", "localId": "claim-006", "confidence": "medium", "status": "argument" }, { "id": "article:audio-first-voice-consumption:claim-007", "type": "claim", "label": "A trigger-based, off-screen audio review layer is a promising near-term pattern, but it should be treated as a testable hypothesis rather than a proven product design.", "localId": "claim-007", "confidence": "medium", "status": "argument" }, { "id": "article:before-machines", "type": "article", "label": "Before Machines: Calculation, Automata, and the Dream of Mechanical Reason", "path": "/articles/before-machines/", "maturity": "seed" }, { "id": "article:before-machines:claim-001", "type": "claim", "label": "Computation was performed by people before it became associated with electronic machines.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:before-machines:claim-002", "type": "claim", "label": "Counting boards and abaci moved arithmetic into visible physical state that could be manipulated and checked.", "localId": "claim-002", "confidence": "medium-high", "status": "argument" }, { "id": "article:before-machines:claim-003", "type": "claim", "label": "Rods, tables, and logarithmic methods reduced complex calculations by decomposing or reusing prior work.", "localId": "claim-003", "confidence": "medium", "status": "argument" }, { "id": "article:before-machines:claim-004", "type": "claim", "label": "Seventeenth-century mechanical calculators embodied arithmetic operations in physical mechanisms.", "localId": "claim-004", "confidence": "high", "status": "argument" }, { "id": "article:before-machines:claim-005", "type": "claim", "label": "The Antikythera mechanism was a sophisticated geared astronomical calculator or display mechanism.", "localId": "claim-005", "confidence": "high", "status": "argument" }, { "id": "article:before-machines:claim-006", "type": "claim", "label": "Automata made mechanism appear self-directed, inviting audiences to project life or agency onto fixed motion.", "localId": "claim-006", "confidence": "medium", "status": "argument" }, { "id": "article:before-machines:claim-007", "type": "claim", "label": "Jacquard punched cards controlled textile patterns and influenced later ideas about machine input and control.", "localId": "claim-007", "confidence": "medium-high", "status": "argument" }, { "id": "article:before-machines:claim-008", "type": "claim", "label": "Babbage's engines mark a threshold between mechanical arithmetic and designs for automatic computing machinery.", "localId": "claim-008", "confidence": "medium-high", "status": "core" }, { "id": "article:birth-of-ai", "type": "article", "label": "The Birth of AI: Dartmouth, Symbolic Systems, and Early Optimism", "path": "/articles/birth-of-ai/", "maturity": "seed" }, { "id": "article:birth-of-ai:claim-001", "type": "claim", "label": "Dartmouth named and consolidated AI as a research field, but did not originate all machine-intelligence work.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:birth-of-ai:claim-002", "type": "claim", "label": "Early AI treated reasoning as symbolic manipulation and search.", "localId": "claim-002", "confidence": "high", "status": "framing" }, { "id": "article:birth-of-ai:claim-003", "type": "claim", "label": "Early demonstrations were impressive but bounded: they worked inside formal or carefully prepared worlds.", "localId": "claim-003", "confidence": "medium-high", "status": "argument" }, { "id": "article:birth-of-ai:claim-004", "type": "claim", "label": "Early AI optimism was part technical, part institutional, and part public narrative.", "localId": "claim-004", "confidence": "medium", "status": "landscape" }, { "id": "article:birth-of-ai:claim-005", "type": "claim", "label": "The early field included multiple lineages, including symbolic reasoning, cybernetics, neural approaches, game-playing, and machine-intelligence philosophy.", "localId": "claim-005", "confidence": "medium-high", "status": "argument" }, { "id": "article:context-management", "type": "article", "label": "Context Management: What the AI Sees Right Now", "path": "/articles/context-management/", "maturity": "seed" }, { "id": "article:context-management:claim-001", "type": "claim", "label": "A model can only work with the information currently in its context window; context management decides what that information is.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:context-management:claim-002", "type": "claim", "label": "Context management resembles human working memory and attention, but it uses fixed-size, lossy windows rather than flexible human recall.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:context-management:claim-003", "type": "claim", "label": "Retrieval and summarization can extend the effective context, but they trade completeness, accuracy, and cost.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:context-management:claim-004", "type": "claim", "label": "Good context management requires deciding what to include, what to compress, and when to stop, because a bigger window is not always a better answer.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:evaluations", "type": "article", "label": "Evaluations: How We Know an AI Workflow Improved", "path": "/articles/evaluations/", "maturity": "seed" }, { "id": "article:evaluations:claim-001", "type": "claim", "label": "An evaluation is a test that turns a quality claim into a repeatable, observable result.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:evaluations:claim-002", "type": "claim", "label": "Benchmarks, report cards, and clinical trials all evaluate outcomes against a standard; AI evaluation extends the same idea to generated outputs and workflows.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:evaluations:claim-003", "type": "claim", "label": "A practical AI evaluation usually mixes automatic checks, human judgments, and task-specific metrics rather than relying on a single score.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:evaluations:claim-004", "type": "claim", "label": "A high score on a benchmark can hide failure modes that matter in real use, because no metric captures every kind of usefulness or harm.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:fine-tuning", "type": "article", "label": "Fine-Tuning: Teaching a Model a Narrower Behavior", "path": "/articles/fine-tuning/", "maturity": "seed" }, { "id": "article:fine-tuning:claim-001", "type": "claim", "label": "Fine-tuning changes a model's learned behavior by continuing training on targeted examples, rather than changing what the model sees at runtime.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:fine-tuning:claim-002", "type": "claim", "label": "Fine-tuning is a form of transfer learning: it adapts a general model to a narrower task using additional examples.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:fine-tuning:claim-003", "type": "claim", "label": "Fine-tuning works best when the task is narrow, the desired outputs are consistent, and high-quality labeled examples are available.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:fine-tuning:claim-004", "type": "claim", "label": "Fine-tuning can bake in errors, biases, or brittle patterns from the training data, so it must be paired with evaluation and clear limits.", "localId": "claim-004", "confidence": "medium-high", "status": "risk" }, { "id": "article:formal-logic-to-computation", "type": "article", "label": "From Formal Logic to Computation: The Mathematical Road to AI", "path": "/articles/formal-logic-to-computation/", "maturity": "seed" }, { "id": "article:formal-logic-to-computation:claim-001", "type": "claim", "label": "Algebraic and symbolic treatments of logic helped make reasoning inspectable and manipulable as formal symbol systems.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:formal-logic-to-computation:claim-003", "type": "claim", "label": "The formalist ambition around mathematical foundations and decision procedures created the problem setting in which computability could be made precise.", "localId": "claim-003", "confidence": "medium-high", "status": "argument" }, { "id": "article:formal-logic-to-computation:claim-004", "type": "claim", "label": "Gödel's incompleteness theorems showed that consistent formal systems strong enough for arithmetic have intrinsic limits, complicating the dream of complete formal foundations.", "localId": "claim-004", "confidence": "high", "status": "landscape" }, { "id": "article:formal-logic-to-computation:claim-005", "type": "claim", "label": "Church, Turing, and Post offered different formalizations of effective procedure, helping turn computation into a mathematical subject before modern computers were common.", "localId": "claim-005", "confidence": "high", "status": "core" }, { "id": "article:formal-logic-to-computation:claim-006", "type": "claim", "label": "The Church-Turing thesis concerns effective methods and is often misunderstood when treated as a claim about all physical machines or minds.", "localId": "claim-006", "confidence": "high", "status": "framing" }, { "id": "article:formal-logic-to-computation:claim-007", "type": "claim", "label": "Shannon's switching-circuit work connected Boolean algebra to relay and switching circuit design, helping make logic part of digital engineering.", "localId": "claim-007", "confidence": "high", "status": "argument" }, { "id": "article:formal-logic-to-computation:claim-008", "type": "claim", "label": "Shannon's communication theory provided a mathematical treatment of messages, channels, noise, and information, but it is not a theory of semantic meaning.", "localId": "claim-008", "confidence": "high", "status": "framing" }, { "id": "article:formal-logic-to-computation:claim-009", "type": "claim", "label": "Cybernetics supplied a language of feedback, control, and communication for thinking about machines and organisms, but feedback alone is not intelligence.", "localId": "claim-009", "confidence": "medium-high", "status": "argument" }, { "id": "article:foundation-models", "type": "article", "label": "Foundation Models and the Return of General-Purpose AI Systems", "path": "/articles/foundation-models/", "maturity": "seed" }, { "id": "article:foundation-models:claim-001", "type": "claim", "label": "A foundation model is a broadly trained model, generally trained with self-supervision at scale, that can be adapted to many downstream tasks.", "localId": "claim-001", "confidence": "high", "status": "framing" }, { "id": "article:foundation-models:claim-002", "type": "claim", "label": "Foundation models revive general-purpose AI ambition by supporting many tasks from a shared base, but this should not be equated with humanlike understanding.", "localId": "claim-002", "confidence": "high", "status": "core" }, { "id": "article:foundation-models:claim-003", "type": "claim", "label": "The Transformer replaced recurrence and convolution with attention for sequence transduction and made training more parallelizable.", "localId": "claim-003", "confidence": "high", "status": "argument" }, { "id": "article:foundation-models:claim-004", "type": "claim", "label": "Broad pretraining enabled models such as BERT and GPT-3 to be adapted or prompted across many tasks.", "localId": "claim-004", "confidence": "high", "status": "argument" }, { "id": "article:foundation-models:claim-005", "type": "claim", "label": "Scaling research made model size, data, and compute explicit variables, while later work emphasized compute-optimal allocation rather than model size alone.", "localId": "claim-005", "confidence": "high", "status": "argument" }, { "id": "article:foundation-models:claim-006", "type": "claim", "label": "Instruction tuning and RLHF can improve usefulness and intent-following, but do not eliminate mistakes or alignment limits.", "localId": "claim-006", "confidence": "high", "status": "risk" }, { "id": "article:foundation-models:claim-007", "type": "claim", "label": "Natural-language supervision and multimodal training widened foundation-model behavior beyond text-only tasks.", "localId": "claim-007", "confidence": "medium-high", "status": "argument" }, { "id": "article:foundation-models:claim-008", "type": "claim", "label": "Retrieval, tool use, and reasoning/action loops can extend model behavior by connecting models to external sources, APIs, and environments.", "localId": "claim-008", "confidence": "high", "status": "argument" }, { "id": "article:foundation-models:claim-009", "type": "claim", "label": "Language-model evaluation needs multi-metric transparency because accuracy alone hides tradeoffs in calibration, robustness, fairness, bias, toxicity, and efficiency.", "localId": "claim-009", "confidence": "high", "status": "risk" }, { "id": "article:foundation-models:claim-010", "type": "claim", "label": "As of 2026-06-19, the 2026 AI Index reports rapid changes in AI capabilities, adoption, incidents, and responsible-AI measurement gaps.", "localId": "claim-010", "confidence": "medium", "status": "landscape" }, { "id": "article:foundation-models:claim-011", "type": "claim", "label": "As of 2026-06-19, NIST AI 600-1 is the generative AI profile used here for lifecycle risk-management framing.", "localId": "claim-011", "confidence": "medium-high", "status": "landscape" }, { "id": "article:foundation-models:claim-012", "type": "claim", "label": "As of 2026-06-19, European Commission pages state that EU general-purpose AI model rules became effective in August 2025 and that the Code of Practice supports compliance.", "localId": "claim-012", "confidence": "medium", "status": "landscape" }, { "id": "article:human-systems", "type": "article", "label": "The Human Road Through AI: Labor, Institutions, Governance, and Meaning", "path": "/articles/human-systems/", "maturity": "seed" }, { "id": "article:human-systems:claim-001", "type": "claim", "label": "AI systems that appear automatic at the interface still depend on human work, judgment, evaluation, maintenance, governance, and contestation.", "localId": "claim-001", "confidence": "medium", "status": "core" }, { "id": "article:human-systems:claim-002", "type": "claim", "label": "Task exposure to generative AI should not be treated as a direct forecast of job replacement.", "localId": "claim-002", "confidence": "medium", "status": "argument" }, { "id": "article:human-systems:claim-003", "type": "claim", "label": "Governance frameworks and laws are part of the AI system because they assign duties for risk, transparency, oversight, accountability, and redress.", "localId": "claim-003", "confidence": "medium", "status": "landscape" }, { "id": "article:human-systems:claim-004", "type": "claim", "label": "Education, science, and authorship debates about AI are debates about human judgment, evidence, disclosure, and responsibility.", "localId": "claim-004", "confidence": "medium", "status": "argument" }, { "id": "article:human-systems:claim-005", "type": "claim", "label": "AI benefits depend on material access conditions: connectivity, devices, language, skills, compute, affordability, energy systems, and local institutions.", "localId": "claim-005", "confidence": "medium", "status": "landscape" }, { "id": "article:human-systems:claim-006", "type": "claim", "label": "Public trust is a design constraint because adoption depends on people's ability to understand, challenge, and rely on AI-mediated systems.", "localId": "claim-006", "confidence": "medium", "status": "argument" }, { "id": "article:learning-machines", "type": "article", "label": "Learning Machines: Statistics, Neural Networks, and the Data Turn", "path": "/articles/learning-machines/", "maturity": "seed" }, { "id": "article:learning-machines:claim-001", "type": "claim", "label": "The shift from hand-coded rules to learning from examples changed AI by combining statistics, neural networks, datasets, benchmarks, compute, and infrastructure into systems that infer useful patterns rather than only follow explicit instructions.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:learning-machines:claim-002", "type": "claim", "label": "The early AI field framing already included learning as a central feature of intelligence.", "localId": "claim-002", "confidence": "high", "status": "framing" }, { "id": "article:learning-machines:claim-003", "type": "claim", "label": "Samuel's checkers work is an early public example of a program improving through machine-learning procedures rather than relying only on fixed, hand-authored play.", "localId": "claim-003", "confidence": "high", "status": "argument" }, { "id": "article:learning-machines:claim-004", "type": "claim", "label": "Rosenblatt's perceptron framed pattern recognition through adaptive connections and probabilistic analysis, making the idea of a learning machine visible to a broad audience.", "localId": "claim-004", "confidence": "high", "status": "argument" }, { "id": "article:learning-machines:claim-005", "type": "claim", "label": "Minsky and Papert analyzed limitations of perceptron models and helped clarify why simple architectures were insufficient for many interesting tasks.", "localId": "claim-005", "confidence": "medium-high", "status": "argument" }, { "id": "article:learning-machines:claim-006", "type": "claim", "label": "The 1986 Nature paper helped make backpropagation for multilayer networks practically legible to a broad research audience, and gradient-trained convolutional networks were already being used for document recognition by the late 1990s.", "localId": "claim-006", "confidence": "high", "status": "argument" }, { "id": "article:learning-machines:claim-007", "type": "claim", "label": "ImageNet and ILSVRC helped make large labeled datasets and shared benchmarks central infrastructure for computer-vision progress.", "localId": "claim-007", "confidence": "high", "status": "landscape" }, { "id": "article:learning-machines:claim-008", "type": "claim", "label": "AlexNet made the combination of deep networks, ImageNet-scale data, and GPU implementation newly persuasive in 2012, but the result should be read as a convergence of factors rather than proof that compute alone or learned patterns equal understanding.", "localId": "claim-008", "confidence": "high", "status": "argument" }, { "id": "article:long-human-road-to-ai", "type": "article", "label": "The Long Human Road to AI: A Reader’s Guide to Season 1", "path": "/articles/long-human-road-to-ai/", "maturity": "seed" }, { "id": "article:long-human-road-to-ai:claim-001", "type": "claim", "label": "Artificial intelligence is easiest to understand when it is presented as the latest chapter in a much longer human story of extending memory, calculation, communication, measurement, coordination, prediction, and delegation.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:long-human-road-to-ai:claim-002", "type": "claim", "label": "Computers and AI emerged from long-running human needs and older tools rather than arriving as a single invention.", "localId": "claim-002", "confidence": "high", "status": "argument" }, { "id": "article:long-human-road-to-ai:claim-003", "type": "claim", "label": "Across the season, the same pattern appears: human need → external support → formalization → scale → boundary.", "localId": "claim-003", "confidence": "medium", "status": "framing" }, { "id": "article:long-human-road-to-ai:claim-004", "type": "claim", "label": "The 1956 Dartmouth workshop named and helped launch AI as a research agenda, but it was one meeting point among many precursors.", "localId": "claim-004", "confidence": "high", "status": "landscape" }, { "id": "article:long-human-road-to-ai:claim-005", "type": "claim", "label": "AI progress repeatedly moved from hand-coded rules and symbols toward learning from examples, then toward scaled general-purpose models.", "localId": "claim-005", "confidence": "medium-high", "status": "argument" }, { "id": "article:long-human-road-to-ai:claim-006", "type": "claim", "label": "Modern AI capabilities are shaped by data, compute, people, organizations, evaluation, governance, infrastructure, and public trust—not only by algorithms.", "localId": "claim-006", "confidence": "medium-high", "status": "risk" }, { "id": "article:long-human-road-to-ai:claim-007", "type": "claim", "label": "Analogies help make AI history understandable, but they are teaching devices, not evidence, and every analogy has a limit.", "localId": "claim-007", "confidence": "high", "status": "framing" }, { "id": "article:long-running-sessions", "type": "article", "label": "Long-Running Sessions: Keeping AI Work Coherent Over Time", "path": "/articles/long-running-sessions/", "maturity": "seed" }, { "id": "article:long-running-sessions:claim-001", "type": "claim", "label": "A long-running session is useful only when the system can remember what matters, recognize progress, and decide when to stop.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:long-running-sessions:claim-002", "type": "claim", "label": "Keeping extended work coherent is already familiar from workflow orchestration, durable execution, project management, and process control.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:long-running-sessions:claim-003", "type": "claim", "label": "In practice, long-running sessions combine summarization, checkpoints, context pruning, and prompt caching to keep the active window focused without losing the goal.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:long-running-sessions:claim-004", "type": "claim", "label": "Without summaries, checkpoints, and stopping rules, long-running sessions drift, waste resources, or resume in broken states.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:long-running-sessions:claim-005", "type": "claim", "label": "Prompt caching can cut the cost and latency of repeated context in long sessions, making extended sessions more practical.", "localId": "claim-005", "confidence": "medium-high", "status": "design" }, { "id": "article:loops-vs-goals", "type": "article", "label": "Loops vs Goals: The Difference Between Repetition and Direction in AI Agents", "path": "/articles/loops-vs-goals/", "maturity": "seed" }, { "id": "article:loops-vs-goals:claim-001", "type": "claim", "label": "A loop repeats work; a goal gives the loop direction and a stopping point.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:loops-vs-goals:claim-002", "type": "claim", "label": "The pairing of loops and goals appears in control engineering, reinforcement learning, and the scientific method, not only in recent AI.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:loops-vs-goals:claim-003", "type": "claim", "label": "Practical AI loops repeat steps such as thinking, acting, observing, editing, or searching until a goal is reached or an exit condition fires.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:loops-vs-goals:claim-004", "type": "claim", "label": "Long-running AI sessions need exit conditions and progress checks to avoid drift, runaway work, or hidden goal substitution.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:memory-vs-context", "type": "article", "label": "Memory vs Context: What Should Survive the Conversation?", "path": "/articles/memory-vs-context/", "maturity": "seed" }, { "id": "article:memory-vs-context:claim-001", "type": "claim", "label": "Context is immediate working material; memory is selected information that persists across time and must be deliberately retrieved or stored.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:memory-vs-context:claim-002", "type": "claim", "label": "The split between immediate context and stored memory appears in cognitive psychology, user interfaces, and database design, not only in recent AI.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:memory-vs-context:claim-003", "type": "claim", "label": "Practical AI systems move information between context and memory through summarization, retrieval, and structured storage, and each transfer is a chance to lose or distort meaning.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:memory-vs-context:claim-004", "type": "claim", "label": "Memory is only useful when retrieval is accurate, updates are careful, and forgetting is as deliberate as remembering.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:multi-agent-systems", "type": "article", "label": "Multi-Agent Systems: When More Than One AI Worker Is Involved", "path": "/articles/multi-agent-systems/", "maturity": "seed" }, { "id": "article:multi-agent-systems:claim-001", "type": "claim", "label": "A multi-agent system solves a problem by dividing it among specialized agents and coordinating their work, not by simply running several models in parallel.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:multi-agent-systems:claim-002", "type": "claim", "label": "Multi-agent systems borrow ideas from distributed systems, ensemble methods, and organizational design, not just recent AI research.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:multi-agent-systems:claim-003", "type": "claim", "label": "Common multi-agent patterns include sequential pipelines, manager-and-workers, and debate-and-review, and each pattern carries different coordination risks.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:multi-agent-systems:claim-004", "type": "claim", "label": "Adding agents increases coordination cost, ambiguity, and failure modes; a multi-agent design should be justified by a specific division of labor, not by default.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:planning-and-reflection", "type": "article", "label": "Planning and Reflection: How AI Breaks Down and Revises Work", "path": "/articles/planning-and-reflection/", "maturity": "seed" }, { "id": "article:planning-and-reflection:claim-001", "type": "claim", "label": "Planning breaks a goal into ordered steps before action; reflection checks results against the goal and decides whether to revise the plan.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:planning-and-reflection:claim-002", "type": "claim", "label": "Planning and reflection are rooted in project management, scientific method, and classical AI search, not only in recent language models.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:planning-and-reflection:claim-003", "type": "claim", "label": "In practice, AI planning and reflection appear as upfront plans, iterative plan-revise loops, and step-by-step reasoning with final verification.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:planning-and-reflection:claim-004", "type": "claim", "label": "Reflection in AI agents is most reliable when paired with external checks such as tests, retrieved sources, or human review; self-critique alone can confirm rather than catch errors.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:planning-and-reflection:claim-005", "type": "claim", "label": "A self-harness pattern can turn one-off reflection into reusable verification routines that are proposed by the model and validated against held-out examples.", "localId": "claim-005", "confidence": "medium", "status": "design" }, { "id": "article:prompt-caching", "type": "article", "label": "Prompt Caching: Reusing Stable Context", "path": "/articles/prompt-caching/", "maturity": "seed" }, { "id": "article:prompt-caching:claim-001", "type": "claim", "label": "Prompt caching reuses the unchanged prefix of a prompt so the provider does not have to reprocess it on every call.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:prompt-caching:claim-002", "type": "claim", "label": "Prompt caching is a specialized form of memoization: it stores the result of an expensive computation so later requests can reuse it.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:prompt-caching:claim-003", "type": "claim", "label": "In practice, prompt caching saves the most when a large, stable prefix is sent repeatedly and the variable part stays at the end.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:prompt-caching:claim-004", "type": "claim", "label": "The savings from prompt caching are bounded by which tokens match, the provider's pricing and retention rules, and whether the same prefix is reused often enough to offset cache-write costs.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:prompt-engineering", "type": "article", "label": "Prompt Engineering: Instruction Design, Not Magic Words", "path": "/articles/prompt-engineering/", "maturity": "seed" }, { "id": "article:prompt-engineering:claim-001", "type": "claim", "label": "A prompt is not just a question; it is the designed instruction, context, examples, and constraints that shape what a language model produces.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:prompt-engineering:claim-002", "type": "claim", "label": "Prompt engineering resembles older practices such as clear writing, task design, and human-computer interaction, updated for probabilistic language models.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:prompt-engineering:claim-003", "type": "claim", "label": "Practical prompt engineering uses techniques—such as giving examples, breaking tasks into steps, and defining output formats—to steer model behavior.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:prompt-engineering:claim-004", "type": "claim", "label": "Prompt engineering is a powerful interface tool, but it cannot fix model errors, guarantee truthfulness, or replace evaluation and oversight.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:reasoning-models", "type": "article", "label": "Reasoning Models: Slower Thinking, Better Checks?", "path": "/articles/reasoning-models/", "maturity": "seed" }, { "id": "article:reasoning-models:claim-001", "type": "claim", "label": "Reasoning models improve hard tasks by deliberately spending more computation on explicit intermediate steps before producing a final answer.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:reasoning-models:claim-002", "type": "claim", "label": "Step-by-step problem solving is an old idea; what changed is scale and language-driven search.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:reasoning-models:claim-003", "type": "claim", "label": "In practice, reasoning models expose a longer trace of intermediate reasoning that can be inspected, even if the trace is not always faithful or complete.", "localId": "claim-003", "confidence": "medium", "status": "design" }, { "id": "article:reasoning-models:claim-004", "type": "claim", "label": "The gains from reasoning models are strongest on complex, well-defined tasks and weakest on simple, ambiguous, or human-judgment tasks.", "localId": "claim-004", "confidence": "medium-high", "status": "risk" }, { "id": "article:retrieval-augmented-generation", "type": "article", "label": "Retrieval-Augmented Generation: Looking Things Up Before Answering", "path": "/articles/retrieval-augmented-generation/", "maturity": "seed" }, { "id": "article:retrieval-augmented-generation:claim-001", "type": "claim", "label": "Retrieval-augmented generation gives a language model relevant external material at request time instead of relying only on its training data and the current prompt.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:retrieval-augmented-generation:claim-002", "type": "claim", "label": "RAG builds on older ideas from information retrieval and open-book question answering: search for sources, then use them to answer.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:retrieval-augmented-generation:claim-003", "type": "claim", "label": "A typical RAG pipeline has three stages: indexing documents, retrieving relevant chunks, and generating an answer conditioned on those chunks.", "localId": "claim-003", "confidence": "high", "status": "design" }, { "id": "article:retrieval-augmented-generation:claim-004", "type": "claim", "label": "RAG reduces some kinds of hallucination, but it cannot fix missing, outdated, or misleading source material, and it can introduce new errors by misusing retrieved passages.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "article:tool-use", "type": "article", "label": "Tool Use: When the Model Calls Something Outside Itself", "path": "/articles/tool-use/", "maturity": "seed" }, { "id": "article:tool-use:claim-001", "type": "claim", "label": "Tool use extends a language model by letting it invoke external capabilities it does not itself possess.", "localId": "claim-001", "confidence": "high", "status": "core" }, { "id": "article:tool-use:claim-002", "type": "claim", "label": "Tool use in AI is conceptually similar to delegation, remote procedure calls, and human use of instruments, but it automates the choice of which tool to invoke.", "localId": "claim-002", "confidence": "high", "status": "landscape" }, { "id": "article:tool-use:claim-003", "type": "claim", "label": "Common tool-use patterns include search, code execution, file or database retrieval, and API calls, each with different reliability and risk profiles.", "localId": "claim-003", "confidence": "medium-high", "status": "design" }, { "id": "article:tool-use:claim-004", "type": "claim", "label": "Tool use introduces failure modes of wrong selection, bad arguments, misplaced trust in tool output, and unwanted actions that must be governed by permissions and checks.", "localId": "claim-004", "confidence": "medium", "status": "risk" }, { "id": "source-001", "type": "source", "label": "Shazeer et al., Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer", "url": "https://arxiv.org/abs/1701.06538", "sourceType": "paper" }, { "id": "source-002", "type": "source", "label": "Fedus et al., Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity", "url": "https://arxiv.org/abs/2101.03961", "sourceType": "paper" }, { "id": "source-003", "type": "source", "label": "Dohan et al., Language Model Cascades", "url": "https://arxiv.org/abs/2207.10342", "sourceType": "paper" }, { "id": "source-004", "type": "source", "label": "Chen et al., FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance", "url": "https://arxiv.org/abs/2305.05176", "sourceType": "paper" }, { "id": "source-005", "type": "source", "label": "Ong et al., RouteLLM: Learning to Route LLMs with Preference Data", "url": "https://arxiv.org/abs/2406.18665", "sourceType": "paper" }, { "id": "source-006", "type": "source", "label": "Yao et al., Tree of Thoughts: Deliberate Problem Solving with Large Language Models", "url": "https://arxiv.org/abs/2305.10601", "sourceType": "paper" }, { "id": "source-007", "type": "source", "label": "Yao et al., ReAct: Synergizing Reasoning and Acting in Language Models", "url": "https://arxiv.org/abs/2210.03629", "sourceType": "paper" }, { "id": "source-008", "type": "source", "label": "Wang et al., Mixture-of-Agents Enhances Large Language Model Capabilities", "url": "https://arxiv.org/abs/2406.04692", "sourceType": "paper" }, { "id": "source-009", "type": "source", "label": "Tang et al., Sakana Fugu: Orchestrator Models for Adaptive Agentic Scaffolds", "url": "https://arxiv.org/abs/2606.21228", "sourceType": "paper" }, { "id": "source-010", "type": "source", "label": "Zhang et al., AFlow: Automating Agentic Workflow Generation", "url": "https://arxiv.org/abs/2410.10762", "sourceType": "paper" }, { "id": "source-011", "type": "source", "label": "LiteLLM documentation", "url": "https://docs.litellm.ai/", "sourceType": "documentation" }, { "id": "source-012", "type": "source", "label": "LangGraph documentation (LangChain)", "url": "https://www.langchain.com/langgraph", "sourceType": "documentation" }, { "id": "source-013", "type": "source", "label": "Wang et al., Self-Consistency Improves Chain of Thought Reasoning in Language Models", "url": "https://arxiv.org/abs/2203.11171", "sourceType": "paper" }, { "id": "source-014", "type": "source", "label": "Shinn et al., Reflexion: Self-Reflective Agents with Dynamic Memory", "url": "https://arxiv.org/abs/2303.11366", "sourceType": "paper" }, { "id": "source-015", "type": "source", "label": "Du et al., Improving Factuality and Reasoning in Language Models through Multiagent Debate", "url": "https://arxiv.org/abs/2305.14325", "sourceType": "paper" }, { "id": "source-016", "type": "source", "label": "Jiang et al., LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusion", "url": "https://arxiv.org/abs/2306.02561", "sourceType": "paper" }, { "id": "source-017", "type": "source", "label": "Xu et al., Trinity: Harmonizing Multiple Large Language Models as a Single Mind", "url": "https://arxiv.org/abs/2512.04695", "sourceType": "paper" }, { "id": "source-018", "type": "source", "label": "Nielsen et al., Conductor: Learning to Orchestrate LLM Agents via Reinforcement Learning", "url": "https://arxiv.org/abs/2512.04388", "sourceType": "paper" }, { "id": "source-019", "type": "source", "label": "Hu et al., RouterBench: A Benchmark for Multi-LLM Routing", "url": "https://arxiv.org/abs/2403.12031", "sourceType": "paper" }, { "id": "source-020", "type": "source", "label": "Hu et al., Automated Design of Agentic Systems", "url": "https://arxiv.org/abs/2408.08435", "sourceType": "paper" }, { "id": "source-021", "type": "source", "label": "Yue et al., MASRouter: A Multiplexing LLM Agent Router", "url": "https://arxiv.org/abs/2502.11133", "sourceType": "paper" }, { "id": "source-022", "type": "source", "label": "Li et al., AgentPrune: Reducing Communication Redundancy in Multi-Agent Systems", "url": "https://arxiv.org/abs/2410.02506", "sourceType": "paper" }, { "id": "source-aba-formal-opinion-512", "type": "source", "label": "American Bar Association, Formal Opinion 512: Generative Artificial Intelligence Tools", "url": "https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf", "sourceType": "report" }, { "id": "source-adobe-ai-traffic-2025", "type": "source", "label": "Adobe Analytics: Traffic to U.S. Retail Websites from Generative AI Sources Jumps 1,200 Percent", "url": "https://blog.adobe.com/en/publish/2025/03/17/adobe-analytics-traffic-to-us-retail-websites-from-generative-ai-sources-jumps-1200-percent", "sourceType": "research" }, { "id": "source-adobe-ai-traffic-2026", "type": "source", "label": "AI traffic grows but retail sites lag in AI search visibility", "url": "https://business.adobe.com/blog/ai-traffic-surge-retail-sites-not-machine-readable", "sourceType": "research" }, { "id": "source-agar-lighthill-2020", "type": "source", "label": "What is science for? The Lighthill report on artificial intelligence reinterpreted", "url": "https://doi.org/10.1017/S0007087420000230", "sourceType": "historical-analysis" }, { "id": "source-agentic-publications-2025", "type": "source", "label": "Agentic Publications: An LLM-Driven Framework for Interactive Scientific Publishing", "url": "https://arxiv.org/abs/2505.13246", "sourceType": "research" }, { "id": "source-alexnet-2012", "type": "source", "label": "ImageNet Classification with Deep Convolutional Neural Networks", "url": "https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks", "sourceType": "research-paper" }, { "id": "source-alpac-1966", "type": "source", "label": "Language and Machines: Computers in Translation and Linguistics", "url": "https://www.nationalacademies.org/read/9547", "sourceType": "public-report" }, { "id": "source-amazon-brand-protection-2024", "type": "source", "label": "How Amazon uses AI innovations to stop fraud and counterfeits", "url": "https://www.aboutamazon.com/news/policy-new-views/amazon-brand-protection-report-2024-counterfeit-products", "sourceType": "industry-report" }, { "id": "source-amazon-rufus", "type": "source", "label": "Amazon Rufus AI experience comes to the Amazon Shopping app", "url": "https://www.aboutamazon.com/news/retail/amazon-rufus", "sourceType": "product" }, { "id": "source-amershi-human-ai-guidelines", "type": "source", "label": "Amershi et al., Guidelines for Human-AI Interaction", "url": "https://www.microsoft.com/en-us/research/wp-content/uploads/2019/01/Guidelines-for-Human-AI-Interaction-camera-ready.pdf", "sourceType": "paper" }, { "id": "source-anthropic-building-effective-agents", "type": "source", "label": "Anthropic, Building effective agents", "url": "https://www.anthropic.com/research/building-effective-agents", "sourceType": "article" }, { "id": "source-anthropic-prompt-caching", "type": "source", "label": "Anthropic: Prompt Caching", "url": "https://docs.anthropic.com/en/docs/build-with-claude/prompt-caching", "sourceType": "documentation" }, { "id": "source-anthropic-prompting", "type": "source", "label": "Anthropic: Prompt engineering overview", "url": "https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview", "sourceType": "product" }, { "id": "source-attention-is-all-you-need", "type": "source", "label": "Attention Is All You Need", "url": "https://arxiv.org/abs/1706.03762", "sourceType": "paper" }, { "id": "source-autogen", "type": "source", "label": "Wu, Bansal, Jin, et al.: AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation", "url": "https://arxiv.org/abs/2308.08155", "sourceType": "paper" }, { "id": "source-autonomous-agents-llm", "type": "source", "label": "Large Language Model based Multi-Agents: A Survey of Progress and Challenges", "url": "https://arxiv.org/abs/2402.01680", "sourceType": "paper" }, { "id": "source-backprop-1986", "type": "source", "label": "Learning Representations by Back-Propagating Errors", "url": "https://doi.org/10.1038/323533a0", "sourceType": "research-paper" }, { "id": "source-baddeley-hitch", "type": "source", "label": "Baddeley & Hitch: Working Memory", "url": "https://www.simplypsychology.org/working-memory.html", "sourceType": "article" }, { "id": "source-bainbridge-ironies-automation", "type": "source", "label": "Bainbridge, Ironies of Automation", "url": "https://ckrybus.com/static/papers/Bainbridge_1983_Automatica.pdf", "sourceType": "paper" }, { "id": "source-bayesian-truth-serum", "type": "source", "label": "Bayesian Truth Serum", "url": "https://nel.mit.edu/bayesian-truth-serum/", "sourceType": "research" }, { "id": "source-beehiiv-mcp-v2", "type": "source", "label": "beehiiv MCP v2", "url": "https://product.beehiiv.com/p/beehiiv-mcp-v2", "sourceType": "product" }, { "id": "source-bert-2018", "type": "source", "label": "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding", "url": "https://arxiv.org/abs/1810.04805", "sourceType": "research" }, { "id": "source-big-bench", "type": "source", "label": "Beyond the Imitation Game Benchmark (BIG-bench)", "url": "https://arxiv.org/abs/2206.04615", "sourceType": "paper" }, { "id": "source-boole-laws-thought", "type": "source", "label": "An Investigation of the Laws of Thought", "url": "https://archive.org/details/investigationofl01bool", "sourceType": "book" }, { "id": "source-britannica-automaton", "type": "source", "label": "Automaton", "url": "https://www.britannica.com/technology/automaton", "sourceType": "encyclopedia" }, { "id": "source-britannica-cybernetics", "type": "source", "label": "Cybernetics", "url": "https://www.britannica.com/science/cybernetics", "sourceType": "documentation" }, { "id": "source-brown-few-shot", "type": "source", "label": "Brown et al.: Language Models are Few-Shot Learners", "url": "https://arxiv.org/abs/2005.14165", "sourceType": "paper" }, { "id": "source-bsharat-instructions", "type": "source", "label": "Bsharat et al.: Principled Instructions Are All You Need for Questioning LLaMA-1/2, GPT-3.5/4", "url": "https://arxiv.org/abs/2312.16171", "sourceType": "paper" }, { "id": "source-c2pa", "type": "source", "label": "C2PA Content Credentials and Provenance Standard", "url": "https://c2pa.org/", "sourceType": "standard" }, { "id": "source-chatgpt-overview", "type": "source", "label": "OpenAI: What can ChatGPT do?", "url": "https://openai.com/chatgpt/overview/", "sourceType": "product" }, { "id": "source-chinchilla-2022", "type": "source", "label": "Training Compute-Optimal Large Language Models", "url": "https://arxiv.org/abs/2203.15556", "sourceType": "research" }, { "id": "source-chm-abacus", "type": "source", "label": "The Versatile, Venerable Abacus", "url": "https://www.computerhistory.org/revolution/calculators/1/1", "sourceType": "museum-exhibit" }, { "id": "source-chm-ai-robotics", "type": "source", "label": "AI and Robotics: Timeline of Computer History", "url": "https://www.computerhistory.org/timeline/ai-robotics/", "sourceType": "article" }, { "id": "source-chm-babbage", "type": "source", "label": "The Babbage Engine", "url": "https://www.computerhistory.org/babbage/", "sourceType": "museum-exhibit" }, { "id": "source-chm-digital-logic", "type": "source", "label": "How Do Digital Computers Think?", "url": "https://www.computerhistory.org/revolution/digital-logic/12/269", "sourceType": "documentation" }, { "id": "source-chm-human-computers", "type": "source", "label": "Human Computers", "url": "https://www.computerhistory.org/revolution/calculators/1/65", "sourceType": "museum-exhibit" }, { "id": "source-chm-leibniz", "type": "source", "label": "The Leibniz Step Reckoner and Curta Calculators", "url": "https://www.computerhistory.org/revolution/calculators/1/49", "sourceType": "museum-exhibit" }, { "id": "source-chm-mech-turk", "type": "source", "label": "Chatbots Decoded: Exploring AI", "url": "https://computerhistory.org/stories/chatbots-decoded/", "sourceType": "museum-story" }, { "id": "source-chm-timeline", "type": "source", "label": "Timeline of Computer History", "url": "https://www.computerhistory.org/timeline/", "sourceType": "museum-timeline" }, { "id": "source-choice-overload", "type": "source", "label": "When Choice is Demotivating: Can One Desire Too Much of a Good Thing?", "url": "https://business.columbia.edu/faculty/research/when-choice-demotivating-can-one-desire-too-much-good-thing", "sourceType": "research" }, { "id": "source-church", "type": "source", "label": "Alonzo Church", "url": "https://plato.stanford.edu/entries/church/", "sourceType": "documentation" }, { "id": "source-church-turing", "type": "source", "label": "The Church-Turing Thesis", "url": "https://plato.stanford.edu/archives/fall2020/entries/church-turing/", "sourceType": "documentation" }, { "id": "source-claude-capabilities", "type": "source", "label": "Anthropic: Claude capabilities", "url": "https://www.anthropic.com/claude", "sourceType": "product" }, { "id": "source-clip-2021", "type": "source", "label": "Learning Transferable Visual Models From Natural Language Supervision", "url": "https://arxiv.org/abs/2103.00020", "sourceType": "research" }, { "id": "source-cloudflare-emdash", "type": "source", "label": "Introducing EmDash", "url": "https://blog.cloudflare.com/emdash-wordpress/", "sourceType": "product" }, { "id": "source-cornell-perceptron", "type": "source", "label": "Professor's perceptron paved the way for AI - 60 years too soon", "url": "https://news.cornell.edu/stories/2019/09/professors-perceptron-paved-way-ai-60-years-too-soon", "sourceType": "article" }, { "id": "source-cot-paper", "type": "source", "label": "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models", "url": "https://arxiv.org/abs/2201.11903", "sourceType": "paper" }, { "id": "source-cowan-working-memory", "type": "source", "label": "Cowan: The Magical Number 4 in Short-Term Memory", "url": "https://doi.org/10.1017/S0140525X01003922", "sourceType": "paper" }, { "id": "source-cpsc-recalls-api", "type": "source", "label": "CPSC Recalls Application Program Interface API Information", "url": "https://www.cpsc.gov/Recalls/CPSC-Recalls-Application-Program-Interface-API-Information", "sourceType": "regulatory-data" }, { "id": "source-dartmouth-1955", "type": "source", "label": "A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence", "url": "https://doi.org/10.1609/aimag.v27i4.1904", "sourceType": "paper" }, { "id": "source-dartmouth-ai-coined", "type": "source", "label": "Artificial Intelligence (AI) Coined at Dartmouth", "url": "https://home.dartmouth.edu/about/artificial-intelligence-ai-coined-dartmouth", "sourceType": "article" }, { "id": "source-dartmouth-story", "type": "source", "label": "AI at Dartmouth: Our Story", "url": "https://ai.dartmouth.edu/our-story", "sourceType": "institutional-history" }, { "id": "source-deep-learning-book", "type": "source", "label": "Deep Learning", "url": "https://www.deeplearningbook.org/", "sourceType": "textbook" }, { "id": "source-deloitte-cpo-2025", "type": "source", "label": "2025 Global Chief Procurement Officer Survey", "url": "https://www.deloitte.com/us/en/about/press-room/2025-chief-procurement-officer-survey.html", "sourceType": "research" }, { "id": "source-dick-ai-history", "type": "source", "label": "Artificial Intelligence", "url": "https://hdsr.mitpress.mit.edu/pub/0aytgrau/release/3", "sourceType": "article" }, { "id": "source-digital-gardens-appleton", "type": "source", "label": "A Brief History & Ethos of the Digital Garden", "url": "https://maggieappleton.com/nontechnical-gardening", "sourceType": "design" }, { "id": "source-digital-gardens-qu", "type": "source", "label": "Digital Gardens", "url": "https://vivqu.com/blog/2020/10/18/digital-gardens/", "sourceType": "design" }, { "id": "source-dpr-karpukhin-2020", "type": "source", "label": "Dense Passage Retrieval for Open-Domain Question Answering", "url": "https://arxiv.org/abs/2004.04906", "sourceType": "paper" }, { "id": "source-drqa-chen-2017", "type": "source", "label": "Reading Wikipedia to Answer Open-Domain Questions", "url": "https://arxiv.org/abs/1704.00051", "sourceType": "paper" }, { "id": "source-edvac-first-draft", "type": "source", "label": "First Draft of a Report on the EDVAC", "url": "https://web.mit.edu/STS.035/www/PDFs/edvac.pdf", "sourceType": "report" }, { "id": "source-eigentrust", "type": "source", "label": "The EigenTrust Algorithm for Reputation Management in P2P Networks", "url": "https://nlp.stanford.edu/pubs/eigentrust.pdf", "sourceType": "research" }, { "id": "source-elements-stat-learning", "type": "source", "label": "The Elements of Statistical Learning", "url": "https://hastie.su.domains/ElemStatLearn/", "sourceType": "textbook" }, { "id": "source-endsley-situation-awareness", "type": "source", "label": "Endsley, Toward a Theory of Situation Awareness in Dynamic Systems", "url": "https://journals.sagepub.com/doi/10.1518/001872095779049543", "sourceType": "paper" }, { "id": "source-epa-safer-choice-products", "type": "source", "label": "Search Products that Meet the Safer Choice Standard", "url": "https://www.epa.gov/saferchoice/products", "sourceType": "regulatory-data" }, { "id": "source-eu-ai-act", "type": "source", "label": "AI Act", "url": "https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai", "sourceType": "policy" }, { "id": "source-eu-ai-act-2024-1689", "type": "source", "label": "EU AI Act, Regulation (EU) 2024/1689", "url": "https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng", "sourceType": "standard" }, { "id": "source-eu-ai-code", "type": "source", "label": "Drawing-up a General-Purpose AI Code of Practice", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-code-practice", "sourceType": "policy" }, { "id": "source-eu-digital-product-passport", "type": "source", "label": "EU's Digital Product Passport: Advancing transparency and sustainability", "url": "https://data.europa.eu/en/news-events/news/eus-digital-product-passport-advancing-transparency-and-sustainability", "sourceType": "regulatory" }, { "id": "source-eval-survey", "type": "source", "label": "Evaluating Large Language Models: A Comprehensive Survey", "url": "https://arxiv.org/abs/2310.19736", "sourceType": "paper" }, { "id": "source-feigenbaum-1977", "type": "source", "label": "The Art of Artificial Intelligence: I. Themes and Case Studies of Knowledge Engineering", "url": "https://stacks.stanford.edu/file/druid%3Abg342cm2034/bg342cm2034.pdf", "sourceType": "conference-paper" }, { "id": "source-feigenbaum-acm", "type": "source", "label": "Edward A. Feigenbaum: A.M. Turing Award Laureate profile", "url": "https://amturing.acm.org/award_winners/feigenbaum_4167235.cfm", "sourceType": "biographical-reference" }, { "id": "source-forrester-b2b-genai-2024", "type": "source", "label": "The Future Of B2B Buying Will Come Slowly And Then All At Once", "url": "https://www.forrester.com/blogs/the-future-of-b2b-buying-will-come-slowly-and-then-all-at-once/", "sourceType": "research" }, { "id": "source-foundation-models-2021", "type": "source", "label": "On the Opportunities and Risks of Foundation Models", "url": "https://arxiv.org/abs/2108.07258", "sourceType": "research-report" }, { "id": "source-frege-logic", "type": "source", "label": "Frege's Logic", "url": "https://plato.stanford.edu/entries/frege-logic/", "sourceType": "documentation" }, { "id": "source-ftc-fake-reviews-2024", "type": "source", "label": "Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials", "url": "https://www.ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commission-announces-final-rule-banning-fake-reviews-testimonials", "sourceType": "regulatory" }, { "id": "source-future-ai-path-automation", "type": "source", "label": "Future AI Path: 10 Everyday Tasks You Can Automate with AI Right Now", "url": "https://futureaipath.com/ai-productivity/task-automation/10-everyday-tasks-you-can-automate-with-ai-right-now/", "sourceType": "guide" }, { "id": "source-gartner-b2b-ai-2026", "type": "source", "label": "Gartner Sales Survey Finds 67% of B2B Buyers Prefer a Rep-Free Experience", "url": "https://www.gartner.com/en/newsroom/press-releases/2026-03-09-gartner-sales-survey-finds-67-percent-of-b2b-buyers-prefer-a-rep-free-experience", "sourceType": "research" }, { "id": "source-gemini-overview", "type": "source", "label": "Google: Gemini overview", "url": "https://gemini.google.com/", "sourceType": "product" }, { "id": "source-generative-agents", "type": "source", "label": "Generative Agents: Interactive Simulacra of Human Behavior", "url": "https://arxiv.org/abs/2304.03442", "sourceType": "paper" }, { "id": "source-gibson-affordances", "type": "source", "label": "Gibson: The Ecological Approach to Visual Perception", "url": "https://www.hachettebookgroup.com/titles/james-j-gibson/the-ecological-approach-to-visual-perception/9780205238818/", "sourceType": "book" }, { "id": "source-godel", "type": "source", "label": "Gödel's Incompleteness Theorems", "url": "https://plato.stanford.edu/entries/goedel-incompleteness/", "sourceType": "documentation" }, { "id": "source-goodfellow-deep-learning", "type": "source", "label": "Goodfellow, Bengio, and Courville: Deep Learning", "url": "https://www.deeplearningbook.org/", "sourceType": "book" }, { "id": "source-google-ap2", "type": "source", "label": "Announcing Agent Payments Protocol (AP2)", "url": "https://cloud.google.com/blog/products/ai-machine-learning/announcing-agents-to-payments-ap2-protocol", "sourceType": "protocol" }, { "id": "source-google-product-structured-data", "type": "source", "label": "Product snippet structured data", "url": "https://developers.google.com/search/docs/appearance/structured-data/product-snippet", "sourceType": "documentation" }, { "id": "source-google-ucp", "type": "source", "label": "Getting started with Universal Commerce Protocol on Google", "url": "https://developers.google.com/merchant/ucp", "sourceType": "protocol" }, { "id": "source-gps-stanford", "type": "source", "label": "A Guide to the General Problem-Solver Program GPS-2-2", "url": "https://stacks.stanford.edu/file/druid:zk239tp3547/zk239tp3547.pdf", "sourceType": "report" }, { "id": "source-gpt3-2020", "type": "source", "label": "Language Models are Few-Shot Learners", "url": "https://arxiv.org/abs/2005.14165", "sourceType": "research" }, { "id": "source-gpt4-2023", "type": "source", "label": "GPT-4 Technical Report", "url": "https://arxiv.org/abs/2303.08774", "sourceType": "technical-report" }, { "id": "source-groundy-chinese-models", "type": "source", "label": "Groundy: Chinese AI Models Compared: DeepSeek, Qwen, Kimi, Doubao, and Ernie", "url": "https://groundy.com/articles/the-chinese-ai-model-ecosystem-deepseek-qwen-kimi-doubao-and-ernie-compared/", "sourceType": "comparison" }, { "id": "source-gs1-digital-link", "type": "source", "label": "GS1 Digital Link for Brand Owners", "url": "https://www.gs1us.org/industries-and-insights/gs1-digital-link/for-brands", "sourceType": "standard" }, { "id": "source-gutwin-greenberg-workspace-awareness", "type": "source", "label": "Gutwin and Greenberg, A Descriptive Framework of Workspace Awareness for Real-Time Groupware", "url": "https://link.springer.com/article/10.1023/A%3A1021271517844", "sourceType": "paper" }, { "id": "source-hai-ai-index", "type": "source", "label": "The 2026 AI Index Report", "url": "https://hai.stanford.edu/ai-index/2026-ai-index-report", "sourceType": "data-reference" }, { "id": "source-haigh-ai-boom-bust", "type": "source", "label": "How the AI Boom Went Bust", "url": "https://cacm.acm.org/opinion/how-the-ai-boom-went-bust/", "sourceType": "historiographic-commentary" }, { "id": "source-haigh-no-first-winter", "type": "source", "label": "There Was No 'First AI Winter'", "url": "https://cacm.acm.org/opinion/there-was-no-first-ai-winter/", "sourceType": "historiographic-commentary" }, { "id": "source-hathway-ai-safety", "type": "source", "label": "Hathway: How to Use AI Safely: 10 Mistakes to Avoid for Secure Usage", "url": "https://www.hathway.com/About/Blog/how-to-use-ai-safely", "sourceType": "safety-guide" }, { "id": "source-helm", "type": "source", "label": "Holistic Evaluation of Language Models (HELM)", "url": "https://arxiv.org/abs/2211.09110", "sourceType": "paper" }, { "id": "source-helm-2022", "type": "source", "label": "Holistic Evaluation of Language Models", "url": "https://arxiv.org/abs/2211.09110", "sourceType": "research" }, { "id": "source-hilbert-problems", "type": "source", "label": "Mathematical Problems", "url": "https://aleph0.clarku.edu/~djoyce/hilbert/problems.html", "sourceType": "documentation" }, { "id": "source-hollan-hutchins-kirsh-distributed-cognition", "type": "source", "label": "Hollan, Hutchins, and Kirsh, Distributed Cognition", "url": "https://www.lri.fr/~mbl/Stanford/CS477/papers/DistributedCognition-TOCHI.pdf", "sourceType": "paper" }, { "id": "source-hong-kong-privacy-tips", "type": "source", "label": "Privacy Commissioner for Personal Data, Hong Kong: 10 Tips for Users of AI Chatbots", "url": "https://www.pcpd.org.hk/english/news_events/media_statements/press_20230913.html", "sourceType": "safety-guide" }, { "id": "source-huggingface-fine-tuning-guide", "type": "source", "label": "Hugging Face: Fine-tuning", "url": "https://huggingface.co/docs/transformers/training", "sourceType": "article" }, { "id": "source-human-in-loop", "type": "source", "label": "Human-in-the-loop Machine Learning: a state of the art", "url": "https://arxiv.org/abs/2108.00981", "sourceType": "paper" }, { "id": "source-iea-energy-ai", "type": "source", "label": "Key Questions on Energy and AI", "url": "https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary", "sourceType": "energy-analysis" }, { "id": "source-ilo-genai-jobs", "type": "source", "label": "Generative AI and Jobs: A Refined Global Index of Occupational Exposure", "url": "https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure", "sourceType": "working-paper" }, { "id": "source-imagenet-2015", "type": "source", "label": "ImageNet Large Scale Visual Recognition Challenge", "url": "https://arxiv.org/abs/1409.0575", "sourceType": "research-paper" }, { "id": "source-imagenet-site", "type": "source", "label": "ImageNet Large Scale Visual Recognition Challenge", "url": "https://www.image-net.org/challenges/LSVRC/", "sourceType": "benchmark-site" }, { "id": "source-instructgpt-paper", "type": "source", "label": "Training language models to follow instructions with human feedback", "url": "https://arxiv.org/abs/2203.02155", "sourceType": "paper" }, { "id": "source-isaw-antikythera", "type": "source", "label": "The Antikythera Mechanism", "url": "https://isaw.nyu.edu/research/antikythera-mechanism", "sourceType": "university-research-summary" }, { "id": "source-itu-facts-2025", "type": "source", "label": "Measuring Digital Development: Facts and Figures 2025", "url": "https://www.itu.int/hub/publication/d-ind-ict_mdd-2025-3/", "sourceType": "official-statistics" }, { "id": "source-langchain-deepagents", "type": "source", "label": "LangChain: Deep Agents", "url": "https://python.langchain.com/docs/concepts/deepagents/", "sourceType": "documentation" }, { "id": "source-langgraph", "type": "source", "label": "LangGraph documentation", "url": "https://docs.langchain.com/oss/python/langgraph/overview", "sourceType": "documentation" }, { "id": "source-langgraph-hitl", "type": "source", "label": "LangGraph human-in-the-loop documentation", "url": "https://docs.langchain.com/oss/python/langchain/human-in-the-loop", "sourceType": "documentation" }, { "id": "source-lecun-docrec-1998", "type": "source", "label": "Gradient-Based Learning Applied to Document Recognition", "url": "https://doi.org/10.1109/5.726791", "sourceType": "research-paper" }, { "id": "source-lee-see-automation-trust", "type": "source", "label": "Trust in Automation: Designing for Appropriate Reliance", "url": "https://journals.sagepub.com/doi/10.1518/hfes.46.1.50_30392", "sourceType": "research" }, { "id": "source-lee-see-trust-automation", "type": "source", "label": "Lee and See, Trust in Automation", "url": "https://journals.sagepub.com/doi/10.1518/hfes.46.1.50_30392", "sourceType": "paper" }, { "id": "source-leibniz-logic", "type": "source", "label": "Leibniz's Influence on 19th Century Logic", "url": "https://plato.stanford.edu/entries/leibniz-logic-influence/", "sourceType": "documentation" }, { "id": "source-lhra-method", "type": "source", "label": "The Long Human Road to AI Research Method", "url": "https://github.com/aura-knowledge/meta/blob/main/proposals/long-human-road-to-ai/research/README.md", "sourceType": "project-document" }, { "id": "source-lighthill-1973", "type": "source", "label": "Artificial Intelligence: A General Survey", "url": "https://www.chilton-computing.org.uk/inf/literature/reports/lighthill_report/p001.htm", "sourceType": "public-report" }, { "id": "source-llm-plus-p", "type": "source", "label": "LLM+P: Empowering Large Language Models with Optimal Planning Proficiency", "url": "https://arxiv.org/abs/2304.11477", "sourceType": "paper" }, { "id": "source-llms-txt", "type": "source", "label": "llms.txt proposal", "url": "https://github.com/answerdotai/llms-txt", "sourceType": "protocol" }, { "id": "source-logic-theory-machine", "type": "source", "label": "The Logic Theory Machine: A Complex Information Processing System", "url": "https://www.rand.org/pubs/papers/P868.html", "sourceType": "paper" }, { "id": "source-lora-paper", "type": "source", "label": "LoRA: Low-Rank Adaptation of Large Language Models", "url": "https://arxiv.org/abs/2106.09685", "sourceType": "paper" }, { "id": "source-lost-in-the-middle", "type": "source", "label": "Lost in the Middle: How Language Models Use Long Contexts", "url": "https://arxiv.org/abs/2307.03172", "sourceType": "paper" }, { "id": "source-lumer-prompt-caching-agents", "type": "source", "label": "Cross-Provider Evaluation of Prompt Caching in Multi-Turn Research Agents", "url": "https://arxiv.org/abs/2601.06007v2", "sourceType": "paper" }, { "id": "source-malone-crowston", "type": "source", "label": "Malone and Crowston, The Interdisciplinary Study of Coordination", "url": "https://crowston.syr.edu/sites/default/files/acmcs94.pdf", "sourceType": "paper" }, { "id": "source-mark-interrupted-work", "type": "source", "label": "Mark, Gudith, and Klocke, The Cost of Interrupted Work", "url": "https://www.ics.uci.edu/~gmark/chi08-mark.pdf", "sourceType": "paper" }, { "id": "source-mccarthy-common-sense", "type": "source", "label": "Programs with Common Sense", "url": "https://jmc.stanford.edu/articles/mcc59/mcc59.pdf", "sourceType": "paper" }, { "id": "source-mcp-introduction", "type": "source", "label": "Model Context Protocol introduction", "url": "https://modelcontextprotocol.io/docs/getting-started/intro", "sourceType": "protocol" }, { "id": "source-mcp-security-best-practices", "type": "source", "label": "Model Context Protocol security best practices", "url": "https://modelcontextprotocol.io/docs/tutorials/security/security_best_practices", "sourceType": "documentation" }, { "id": "source-memgpt", "type": "source", "label": "Packer et al.: MemGPT: Towards LLMs as Operating Systems", "url": "https://arxiv.org/abs/2310.08560", "sourceType": "paper" }, { "id": "source-minsky-papert-1969", "type": "source", "label": "Perceptrons: An Introduction to Computational Geometry", "url": "https://mitpress.mit.edu/9780262630221/perceptrons/", "sourceType": "book" }, { "id": "source-mmlu", "type": "source", "label": "Measuring Massive Multitask Language Understanding", "url": "https://arxiv.org/abs/2009.03300", "sourceType": "paper" }, { "id": "source-mpiwg-antikythera", "type": "source", "label": "The Inscriptions of the Antikythera Mechanism", "url": "https://www.mpiwg-berlin.mpg.de/research/projects/inscriptions-antikythera-mechanism", "sourceType": "research-project" }, { "id": "source-multi-agent-debate", "type": "source", "label": "Du, Li, Tarlow, et al.: Improving Factuality and Reasoning in Language Models through Multiagent Debate", "url": "https://arxiv.org/abs/2305.14325", "sourceType": "paper" }, { "id": "source-mycin-1984", "type": "source", "label": "Rule-Based Expert Systems: The MYCIN Experiments of the Stanford Heuristic Programming Project", "url": "https://www.shortliffe.net/Buchanan-Shortliffe-1984/MYCIN%20Book.htm", "sourceType": "expert-system-retrospective" }, { "id": "source-nature-antikythera-2006", "type": "source", "label": "Decoding the ancient Greek astronomical calculator known as the Antikythera Mechanism", "url": "https://www.nature.com/articles/nature05357", "sourceType": "research-article" }, { "id": "source-nilsson-quest-ai", "type": "source", "label": "The Quest for Artificial Intelligence: A History of Ideas and Achievements", "url": "https://ai.stanford.edu/~nilsson/QAI/qai.pdf", "sourceType": "book" }, { "id": "source-nist-ai-rmf", "type": "source", "label": "AI Risk Management Framework", "url": "https://www.nist.gov/itl/ai-risk-management-framework", "sourceType": "standard" }, { "id": "source-nist-gai-profile", "type": "source", "label": "NIST AI RMF Generative AI Profile", "url": "https://doi.org/10.6028/NIST.AI.600-1", "sourceType": "guidance" }, { "id": "source-nist-math-tables", "type": "source", "label": "Prehistory: The Math Tables Project", "url": "https://www.nist.gov/mathematics-statistics/prehistory-math-tables-project", "sourceType": "government-history" }, { "id": "source-nng-cognitive-load", "type": "source", "label": "4 Principles for Reducing Cognitive Load in Forms", "url": "https://www.nngroup.com/articles/4-principles-reduce-cognitive-load/", "sourceType": "design" }, { "id": "source-nrc-ai-1999", "type": "source", "label": "Funding a Revolution: Government Support for Computing Research, Chapter 9: Development in Artificial Intelligence", "url": "https://www.nationalacademies.org/read/6323/chapter/11", "sourceType": "institutional-history" }, { "id": "source-oecd-ai-principles", "type": "source", "label": "OECD AI Principles", "url": "https://oecd.ai/en/ai-principles", "sourceType": "intergovernmental-principles" }, { "id": "source-oecd-ai-science", "type": "source", "label": "Artificial Intelligence in Science", "url": "https://doi.org/10.1787/a8d820bd-en", "sourceType": "research-synthesis" }, { "id": "source-openai-agents", "type": "source", "label": "OpenAI Agents guide", "url": "https://developers.openai.com/api/docs/guides/agents", "sourceType": "documentation" }, { "id": "source-openai-agents-hitl", "type": "source", "label": "OpenAI Agents SDK human-in-the-loop documentation", "url": "https://openai.github.io/openai-agents-python/human_in_the_loop/", "sourceType": "documentation" }, { "id": "source-openai-agents-tracing", "type": "source", "label": "OpenAI Agents SDK tracing", "url": "https://openai.github.io/openai-agents-python/tracing/", "sourceType": "documentation" }, { "id": "source-openai-chatgpt-shopping", "type": "source", "label": "Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol", "url": "https://openai.com/index/buy-it-in-chatgpt/", "sourceType": "product" }, { "id": "source-openai-fine-tuning-guide", "type": "source", "label": "OpenAI: Fine-tuning guide", "url": "https://platform.openai.com/docs/guides/fine-tuning", "sourceType": "article" }, { "id": "source-openai-merchants", "type": "source", "label": "Power product discovery in ChatGPT", "url": "https://chatgpt.com/merchants/", "sourceType": "product" }, { "id": "source-openai-o1", "type": "source", "label": "OpenAI: Learning to Reason with LLMs", "url": "https://openai.com/index/learning-to-reason-with-llms/", "sourceType": "article" }, { "id": "source-openai-prompt-caching", "type": "source", "label": "OpenAI: Prompt Caching", "url": "https://platform.openai.com/docs/guides/prompt-caching", "sourceType": "documentation" }, { "id": "source-openai-prompt-guide", "type": "source", "label": "OpenAI: Prompt Engineering", "url": "https://platform.openai.com/docs/guides/prompt-engineering", "sourceType": "documentation" }, { "id": "source-openai-stripe-acp", "type": "source", "label": "Stripe powers Instant Checkout in ChatGPT and releases Agentic Commerce Protocol", "url": "https://stripe.com/newsroom/news/stripe-openai-instant-checkout", "sourceType": "protocol" }, { "id": "source-openai-usage-study", "type": "source", "label": "TechRadar: OpenAI reveals how people use ChatGPT — 3 things we learned", "url": "https://www.techradar.com/ai-platforms-assistants/chatgpt/openai-reveals-biggest-ever-study-of-how-people-are-using-chatgpt-here-are-3-things-weve-learned", "sourceType": "research" }, { "id": "source-openfda-food-enforcement", "type": "source", "label": "openFDA Food Enforcement API", "url": "https://open.fda.gov/apis/food/enforcement/", "sourceType": "regulatory-data" }, { "id": "source-pagedattention", "type": "source", "label": "vLLM: PagedAttention for Efficient LLM Serving", "url": "https://arxiv.org/abs/2309.06180", "sourceType": "paper" }, { "id": "source-pai-data-labor", "type": "source", "label": "Responsible Sourcing Across the Data Supply Line", "url": "https://partnershiponai.org/workstream/responsible-sourcing/", "sourceType": "practice-guidance" }, { "id": "source-pew-ai-topic", "type": "source", "label": "Artificial Intelligence", "url": "https://www.pewresearch.org/topic/internet-technology/emerging-technology/artificial-intelligence/", "sourceType": "survey-topic" }, { "id": "source-post-1936", "type": "source", "label": "Finite Combinatory Processes - Formulation 1", "url": "https://doi.org/10.2307/2269031", "sourceType": "article" }, { "id": "source-principia-archive", "type": "source", "label": "Principia Mathematica", "url": "https://archive.org/details/cu31924001575244", "sourceType": "book" }, { "id": "source-privacy-pass-architecture", "type": "source", "label": "The Privacy Pass Architecture", "url": "https://datatracker.ietf.org/doc/rfc9576/", "sourceType": "standard" }, { "id": "source-provenance-2026", "type": "source", "label": "Responsible Agentic AI Requires Explicit Provenance", "url": "https://arxiv.org/abs/2605.17169", "sourceType": "research" }, { "id": "source-r1-1980", "type": "source", "label": "R1: An Expert in the Computer Systems Domain", "url": "https://cdn.aaai.org/AAAI/1980/AAAI80-076.pdf", "sourceType": "conference-paper" }, { "id": "source-r1-revisited-1984", "type": "source", "label": "R1 Revisited: Four Years in the Trenches", "url": "https://ojs.aaai.org/aimagazine/index.php/aimagazine/article/view/445", "sourceType": "case-study" }, { "id": "source-rag", "type": "source", "label": "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks", "url": "https://arxiv.org/abs/2005.11401", "sourceType": "paper" }, { "id": "source-rag-2020", "type": "source", "label": "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks", "url": "https://arxiv.org/abs/2005.11401", "sourceType": "research" }, { "id": "source-rag-lewis-2020", "type": "source", "label": "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks", "url": "https://arxiv.org/abs/2005.11401", "sourceType": "paper" }, { "id": "source-rag-survey-gao-2024", "type": "source", "label": "Retrieval-Augmented Generation for Large Language Models: A Survey", "url": "https://arxiv.org/abs/2312.10997", "sourceType": "paper" }, { "id": "source-react-2022", "type": "source", "label": "ReAct: Synergizing Reasoning and Acting in Language Models", "url": "https://arxiv.org/abs/2210.03629", "sourceType": "research" }, { "id": "source-react-paper", "type": "source", "label": "ReAct: Synergizing Reasoning and Acting in Language Models", "url": "https://arxiv.org/abs/2210.03629", "sourceType": "paper" }, { "id": "source-reflexion-paper", "type": "source", "label": "Reflexion: Language Agents with Verbal Reinforcement Learning", "url": "https://arxiv.org/abs/2303.11366", "sourceType": "paper" }, { "id": "source-rlhf-2022", "type": "source", "label": "Training language models to follow instructions with human feedback", "url": "https://arxiv.org/abs/2203.02155", "sourceType": "research" }, { "id": "source-rosenblatt-1958", "type": "source", "label": "The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain", "url": "https://homepages.math.uic.edu/~lreyzin/papers/rosenblatt58.pdf", "sourceType": "paper" }, { "id": "source-russell-norvig-aima", "type": "source", "label": "Russell and Norvig: Artificial Intelligence: A Modern Approach", "url": "https://aima.cs.berkeley.edu/", "sourceType": "book" }, { "id": "source-samuel-1959", "type": "source", "label": "Some Studies in Machine Learning Using the Game of Checkers", "url": "https://doi.org/10.1147/rd.33.0210", "sourceType": "research-paper" }, { "id": "source-scaling-laws-2020", "type": "source", "label": "Scaling Laws for Neural Language Models", "url": "https://arxiv.org/abs/2001.08361", "sourceType": "research" }, { "id": "source-schema-product", "type": "source", "label": "Schema.org Product", "url": "https://schema.org/Product", "sourceType": "schema" }, { "id": "source-schema-review", "type": "source", "label": "Schema.org Review", "url": "https://schema.org/Review", "sourceType": "schema" }, { "id": "source-schick-toolformer", "type": "source", "label": "Toolformer: Language Models Can Teach Themselves to Use Tools", "url": "https://arxiv.org/abs/2302.04761", "sourceType": "paper" }, { "id": "source-science-museum-automata", "type": "source", "label": "Marvellous Machines: Early Robots", "url": "https://www.sciencemuseum.org.uk/objects-and-stories/marvellous-machines-early-robots", "sourceType": "museum-story" }, { "id": "source-self-harness-paper", "type": "source", "label": "Self-Harness: Mining Weaknesses to Generate Test-Time Harnesses", "url": "https://arxiv.org/abs/2606.09498", "sourceType": "paper" }, { "id": "source-self-refine-paper", "type": "source", "label": "Self-Refine: Iterative Refinement with Self-Feedback", "url": "https://arxiv.org/abs/2303.17651", "sourceType": "paper" }, { "id": "source-sep-ai", "type": "source", "label": "Artificial Intelligence", "url": "https://plato.stanford.edu/entries/artificial-intelligence/", "sourceType": "article" }, { "id": "source-sep-computing", "type": "source", "label": "The Modern History of Computing", "url": "https://plato.stanford.edu/entries/computing-history/", "sourceType": "encyclopedia" }, { "id": "source-sep-leibniz-logic", "type": "source", "label": "Leibniz's Influence on 19th Century Logic", "url": "https://plato.stanford.edu/entries/leibniz-logic-influence/", "sourceType": "encyclopedia" }, { "id": "source-sep-logic-ai", "type": "source", "label": "Logic and Artificial Intelligence", "url": "https://plato.stanford.edu/entries/logic-ai/", "sourceType": "article" }, { "id": "source-series-proposal", "type": "source", "label": "AI, De-Mystified Series Proposal", "url": "https://github.com/aura-knowledge/meta/blob/main/proposals/ai-demystified/README.md", "sourceType": "article" }, { "id": "source-shannon-info-1948", "type": "source", "label": "A Mathematical Theory of Communication", "url": "https://doi.org/10.1002/j.1538-7305.1948.tb01338.x", "sourceType": "article" }, { "id": "source-shannon-info-pdf", "type": "source", "label": "A Mathematical Theory of Communication", "url": "https://ia803209.us.archive.org/27/items/bstj27-3-379/bstj27-3-379_text.pdf", "sourceType": "article" }, { "id": "source-shannon-switching", "type": "source", "label": "A Symbolic Analysis of Relay and Switching Circuits", "url": "https://doi.org/10.1109/T-AIEE.1938.5057767", "sourceType": "article" }, { "id": "source-simon-attention-scarcity", "type": "source", "label": "Designing Organizations for an Information-Rich World", "url": "https://www.econbiz.de/Record/designing-organizations-for-an-information-rich-world-simon-herbert/10002817747", "sourceType": "research" }, { "id": "source-smg-babbage", "type": "source", "label": "Charles Babbage's Difference Engines and the Science Museum", "url": "https://www.sciencemuseum.org.uk/objects-and-stories/charles-babbages-difference-engines-and-science-museum", "sourceType": "museum-story" }, { "id": "source-smg-jacquard", "type": "source", "label": "Programming Patterns: The Story of the Jacquard Loom", "url": "https://www.scienceandindustrymuseum.org.uk/objects-and-stories/jacquard-loom", "sourceType": "museum-story" }, { "id": "source-smg-pascal-calculator", "type": "source", "label": "Replica of Pascal's Calculator", "url": "https://collection.sciencemuseumgroup.org.uk/objects/co60769/replica-of-pascals-calculator", "sourceType": "museum-object" }, { "id": "source-smith-human-computers", "type": "source", "label": "Human Computer Project", "url": "https://womenshistory.si.edu/human-computer-project", "sourceType": "museum-research-project" }, { "id": "source-smithsonian-abacus", "type": "source", "label": "The Abacus, the Numeral Frame, and Counters", "url": "https://www.si.edu/spotlight/the-abacus-the-numeral-frame-and-counters", "sourceType": "museum-exhibit" }, { "id": "source-smithsonian-napier-rods", "type": "source", "label": "Napier's Rods", "url": "https://americanhistory.si.edu/collections/nmah_690812", "sourceType": "museum-object" }, { "id": "source-smithsonian-punch-cards", "type": "source", "label": "Punch Cards", "url": "https://www.si.edu/spotlight/punch-cards", "sourceType": "museum-exhibit" }, { "id": "source-status-quo-bias", "type": "source", "label": "Status Quo Bias in Decision Making", "url": "https://rzeckhauser.scholars.harvard.edu/publications/status-quo-bias-decision-making", "sourceType": "research" }, { "id": "source-sutton-barto-rl", "type": "source", "label": "Sutton and Barto: Reinforcement Learning: An Introduction", "url": "http://incompleteideas.net/book/the-book-2nd.html", "sourceType": "book" }, { "id": "source-sutton-bitter", "type": "source", "label": "The Bitter Lesson", "url": "https://www.incompleteideas.net/IncIdeas/BitterLesson.html", "sourceType": "essay" }, { "id": "source-tech-insider-models", "type": "source", "label": "Tech Insider: ChatGPT vs Claude vs Gemini vs DeepSeek [2026]", "url": "https://tech-insider.org/chatgpt-vs-claude-vs-deepseek-vs-gemini-2026/", "sourceType": "comparison" }, { "id": "source-toolformer-2023", "type": "source", "label": "Toolformer: Language Models Can Teach Themselves to Use Tools", "url": "https://arxiv.org/abs/2302.04761", "sourceType": "research" }, { "id": "source-toolformer-paper", "type": "source", "label": "Toolformer: Language Models Can Teach Themselves to Use Tools", "url": "https://arxiv.org/abs/2302.04761", "sourceType": "paper" }, { "id": "source-transformer-2017", "type": "source", "label": "Attention Is All You Need", "url": "https://arxiv.org/abs/1706.03762", "sourceType": "research" }, { "id": "source-trinity-ai-privacy", "type": "source", "label": "Trinity College: Security Tip: AI & Data Privacy Best Practices", "url": "https://www.trincoll.edu/lits/technology/security/best-practices/security-tips/ai-data-privacy/", "sourceType": "safety-guide" }, { "id": "source-truth-discovery-survey", "type": "source", "label": "A Survey on Truth Discovery", "url": "https://www.kdd.org/exploration_files/Article1_17_2.pdf", "sourceType": "research" }, { "id": "source-tulving-memory", "type": "source", "label": "Tulving: Episodic and Semantic Memory", "url": "https://plato.stanford.edu/entries/memory/", "sourceType": "article" }, { "id": "source-turing-1936", "type": "source", "label": "On Computable Numbers, with an Application to the Entscheidungsproblem", "url": "https://doi.org/10.1112/plms/s2-42.1.230", "sourceType": "article" }, { "id": "source-turing-1950", "type": "source", "label": "Computing Machinery and Intelligence", "url": "https://turingarchive.kings.cam.ac.uk/computing-machinery-and-intelligence", "sourceType": "article" }, { "id": "source-unesco-genai-edu", "type": "source", "label": "Guidance for Generative AI in Education and Research", "url": "https://unesdoc.unesco.org/ark:/48223/pf0000386693", "sourceType": "official-guidance" }, { "id": "source-unesco-generative-ai-education-research", "type": "source", "label": "UNESCO guidance on generative AI in education and research", "url": "https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research", "sourceType": "report" }, { "id": "source-usco-ai", "type": "source", "label": "Copyright and Artificial Intelligence", "url": "https://www.copyright.gov/ai/", "sourceType": "official-report-series" }, { "id": "source-w3c-bbs", "type": "source", "label": "Data Integrity BBS Cryptosuites v1.0", "url": "https://www.w3.org/TR/vc-di-bbs/", "sourceType": "standard" }, { "id": "source-w3c-prov-o", "type": "source", "label": "PROV-O: The PROV Ontology", "url": "https://www.w3.org/TR/prov-o/", "sourceType": "standard" }, { "id": "source-w3c-vc-data-model", "type": "source", "label": "Verifiable Credentials Data Model v2.0", "url": "https://www.w3.org/TR/vc-data-model-2.0/", "sourceType": "standard" }, { "id": "source-wei-cot", "type": "source", "label": "Wei et al.: Chain-of-Thought Prompting Elicits Reasoning in Large Language Models", "url": "https://arxiv.org/abs/2201.11903", "sourceType": "paper" }, { "id": "source-wei-cot-2022", "type": "source", "label": "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models", "url": "https://arxiv.org/abs/2201.11903", "sourceType": "paper" }, { "id": "source-whipple-napier", "type": "source", "label": "John Napier's Calculating Tools", "url": "https://www.whipplemuseum.cam.ac.uk/explore-whipple-collections/calculating-devices/john-napiers-calculating-tools", "sourceType": "museum-exhibit" }, { "id": "source-wiener-cybernetics", "type": "source", "label": "Cybernetics: Or Control and Communication in the Animal and the Machine", "url": "https://doi.org/10.7551/mitpress/11810.001.0001", "sourceType": "book" }, { "id": "source-wikipedia-memoization", "type": "source", "label": "Wikipedia: Memoization", "url": "https://en.wikipedia.org/wiki/Memoization", "sourceType": "article" }, { "id": "source-wooldridge-mas", "type": "source", "label": "Wooldridge: An Introduction to MultiAgent Systems", "url": "https://www.cs.ox.ac.uk/people/michael.wooldridge/pubs/imas/IMAS2e.pdf", "sourceType": "book" }, { "id": "source-wordpress-mcp-adapter", "type": "source", "label": "From Abilities to AI Agents: Introducing the WordPress MCP Adapter", "url": "https://developer.wordpress.org/news/2026/02/from-abilities-to-ai-agents-introducing-the-wordpress-mcp-adapter/", "sourceType": "product" }, { "id": "source-worldbank-digital-2023", "type": "source", "label": "Digital Progress and Trends Report 2023", "url": "https://doi.org/10.1596/978-1-4648-2049-6", "sourceType": "institutional-report" }, { "id": "source-yao-tot-2023", "type": "source", "label": "Tree of Thoughts: Deliberate Problem Solving with Large Language Models", "url": "https://arxiv.org/abs/2305.10601", "sourceType": "paper" }, { "id": "topic:agent-orchestration", "type": "topic", "label": "agent-orchestration" }, { "id": "topic:agent-provenance", "type": "topic", "label": "agent-provenance" }, { "id": "topic:agentic-commerce", "type": "topic", "label": "agentic-commerce" }, { "id": "topic:ai-agents", "type": "topic", "label": "ai-agents" }, { "id": "topic:ai-governance", "type": "topic", "label": "ai-governance" }, { "id": "topic:ai-history", "type": "topic", "label": "ai-history" }, { "id": "topic:ai-literacy", "type": "topic", "label": "ai-literacy" }, { "id": "topic:ai-native-publishing", "type": "topic", "label": "ai-native-publishing" }, { "id": "topic:ai-society", "type": "topic", "label": "ai-society" }, { "id": "topic:audio", "type": "topic", "label": "audio" }, { "id": "topic:computability", "type": "topic", "label": "computability" }, { "id": "topic:computing-history", "type": "topic", "label": "computing-history" }, { "id": "topic:consumer-behavior", "type": "topic", "label": "consumer-behavior" }, { "id": "topic:cybernetics", "type": "topic", "label": "cybernetics" }, { "id": "topic:digital-gardens", "type": "topic", "label": "digital-gardens" }, { "id": "topic:education", "type": "topic", "label": "education" }, { "id": "topic:evaluation", "type": "topic", "label": "evaluation" }, { "id": "topic:expert-systems", "type": "topic", "label": "expert-systems" }, { "id": "topic:formal-logic", "type": "topic", "label": "formal-logic" }, { "id": "topic:foundation-models", "type": "topic", "label": "foundation-models" }, { "id": "topic:governance", "type": "topic", "label": "governance" }, { "id": "topic:human-ai-interaction", "type": "topic", "label": "human-ai-interaction" }, { "id": "topic:human-progress", "type": "topic", "label": "human-progress" }, { "id": "topic:hype-cycle", "type": "topic", "label": "hype-cycle" }, { "id": "topic:incumbent-adjacent-ventures", "type": "topic", "label": "incumbent-adjacent-ventures" }, { "id": "topic:information-theory", "type": "topic", "label": "information-theory" }, { "id": "topic:institutions", "type": "topic", "label": "institutions" }, { "id": "topic:labor", "type": "topic", "label": "labor" }, { "id": "topic:llm-routing", "type": "topic", "label": "llm-routing" }, { "id": "topic:long-human-road-to-ai", "type": "topic", "label": "long-human-road-to-ai" }, { "id": "topic:machine-learning", "type": "topic", "label": "machine-learning" }, { "id": "topic:neural-networks", "type": "topic", "label": "neural-networks" }, { "id": "topic:onboarding", "type": "topic", "label": "onboarding" }, { "id": "topic:open-protocols", "type": "topic", "label": "open-protocols" }, { "id": "topic:privacy", "type": "topic", "label": "privacy" }, { "id": "topic:product-truth", "type": "topic", "label": "product-truth" }, { "id": "topic:prompting", "type": "topic", "label": "prompting" }, { "id": "topic:research-workflows", "type": "topic", "label": "research-workflows" }, { "id": "topic:scaling", "type": "topic", "label": "scaling" }, { "id": "topic:symbolic-ai", "type": "topic", "label": "symbolic-ai" }, { "id": "topic:transformers", "type": "topic", "label": "transformers" }, { "id": "topic:voice-interfaces", "type": "topic", "label": "voice-interfaces" } ]