SocioLens

Debates

• What kind of thing is generative AI, and does the answer predict anything?
  Researchers, firms, regulators and governments describe generative AI in incompatible terms: a tool, a normal technology, a cultural technology, an agent, a partner in communication, an infrastructure, an investment boom. Which description is best, and does the choice predict anything a researcher could observe?
• Aggregate null or distributional shift? Employment
  What are the early labour-market effects of generative AI?
• Organizational absorption: Why micro gains do not become macro outcomes
  Experiments find that generative AI saves workers time on many tasks. Why does so little of that show up in firm performance and national productivity, at least so far?
• Levelling or hollowing? Skill compression and the reproduction of expertise
  When generative AI helps the least experienced workers most, does it level expertise, or does it hollow out the way expertise is reproduced?
• Discretion curtailed, relocated or encoded?
  When public organisations introduce algorithmic, machine-learning or foundation-model systems into administrative decision-making, does human discretion actually shrink?
• The evaluation apparatus: Can benchmarks, leaderboards and audits carry the weight placed on them?
  Benchmark scores, leaderboard ranks, model judges, developers' own reports, red-teaming and audits are now used to decide what research is published, which models are bought, which are released and which comply with the law. Can they carry that weight?
• AI in the research apparatus
  Under what conditions can foundation-model-mediated social research produce valid knowledge?
• Who can know about AI? Epistemic dependence and jurisdiction over the field
  AI firms build the largest models, hold the compute, collect the records of how people use AI, fund research on its effects and decide which outside researchers may test their systems. How independent can research on AI's social effects be under these conditions, and who is gaining the authority to say what AI does to society?
• Individual gain, collective narrowing? Does AI make science, creative work and shared knowledge more alike?
  AI tools help individual scientists publish and be cited more, help individual writers produce work that readers rate higher, and give individual users quick answers. Do the same tools make the work of many people more alike, narrow what a field studies and thin out the shared stores of knowledge everyone draws on?
• Persuasion, Intermediation & the Information Order
  When, and through which mechanisms, does generative AI change persuasion, information intermediation and the public information environment?
• Concentration, Stack & Sovereignty
  Where is power concentrated across the AI stack, through which mechanisms does that concentration matter, and under what conditions can firms, states, researchers or publics reduce dependency?
• Regulation-in-Practice: Rights, Compliance & Developmental Steering
  What does AI regulation become after legislation is announced or enacted?

Research opportunities: 75 literature-screened directions drawn from the debates' open questions.

Start Here

• START-HERE — SocioLens — Start Here v0.1

Field Map

• SocioLens — Field Map v0.2

Living Debates

• LD-01 — LD-01 — WHAT KIND OF THING IS GENERATIVE AI, AND DOES THE ANSWER PREDICT ANYTHING?
• LD-02 — LD-02 — AGGREGATE NULL OR DISTRIBUTIONAL SHIFT? EMPLOYMENT
• LD-03 — LD-03 — ORGANIZATIONAL ABSORPTION: WHY MICRO GAINS DO NOT BECOME MACRO OUTCOMES
• LD-04 — LD-04 — LEVELLING OR HOLLOWING? SKILL COMPRESSION AND THE REPRODUCTION OF EXPERTISE
• LD-05 — LD-05 — DISCRETION CURTAILED, RELOCATED OR ENCODED?
• LD-06 — LD-06 — THE EVALUATION APPARATUS: CAN BENCHMARKS, LEADERBOARDS AND AUDITS CARRY THE WEIGHT PLACED ON THEM?
• LD-07 — LD-07 — AI IN THE RESEARCH APPARATUS
• LD-08 — LD-08 — WHO CAN KNOW ABOUT AI? EPISTEMIC DEPENDENCE AND JURISDICTION OVER THE FIELD
• LD-09 — LD-09 — INDIVIDUAL GAIN, COLLECTIVE NARROWING? DOES AI MAKE SCIENCE, CREATIVE WORK AND SHARED KNOWLEDGE MORE ALIKE?
• LD-10 — LD-10 — Persuasion, Intermediation & the Information Order
• LD-11 — LD-11 — Concentration, Stack & Sovereignty
• LD-12 — LD-12 — Regulation-in-Practice: Rights, Compliance & Developmental Steering

Claims

• LD01-SL-C01 — Narayanan and Kapoor describe AI as “normal technology” and predict that slow diffusion “will continue to be the norm in
• LD01-SL-C02 — Narayanan and Kapoor predict that AI will not “meaningfully outperform trained humans (particularly teams of humans and
• LD01-SL-C03 — Reading a 2024 study that found 40% of US adults used generative AI, Narayanan and Kapoor note that infrequent use trans
• LD01-SL-C04 — Yiu, Kosoy and Gopnik describe large language models as “cultural technologies” and “efficient imitation engines,” and p
• LD01-SL-C05 — Mitchell and Krakauer report that in a 2022 survey of natural-language-processing researchers, 51% of 480 respondents ag
• LD01-SL-C06 — Shanahan, McDonell and Reynolds argue that a dialogue agent is best understood as role-play by a simulator, not as an en
• LD01-SL-C07 — Millière and Buckner take the view that language models are “merely sophisticated mimics” as a null hypothesis to be tes
• LD01-SL-C08 — Esposito argues that comparing algorithms with human intelligence is misleading and proposes thinking of them “not in te
• LD01-SL-C09 — Drawing on Luhmann, Esposito, Wittgenstein and Brandom, Fenoglio argues that what language models lack is a position in
• LD01-SL-C10 — Chan and colleagues treat agency in algorithmic systems as a matter of degree, raised by four characteristics: underspec
• LD01-SL-C11 — Kolt notes that the US Restatement (Third) of Agency holds that a computer program “is not capable of acting as a princi
• LD01-SL-C12 — Kasirzadeh and Gabriel characterize AI agents along four dimensions, autonomy, efficacy, goal complexity and generality,
• LD01-SL-C13 — Anthropic distinguishes “workflows,” in which language models and tools are orchestrated through predefined code paths,
• LD01-SL-C14 — Eloundou, Manning, Mishkin and Rock conclude that language models “exhibit traits of general-purpose technologies,” esti
• LD01-SL-C15 — In nationally representative US surveys from late 2024, nearly 40% of adults aged 18–64 used generative AI; 27% of worke
• LD01-SL-C16 — A GPT-style model trained only on sequences of legal Othello moves, with no rules or board given, made illegal moves 0.0
• LD01-SL-C17 — Linear probes on Llama-2-70B’s activations predicted the locations of world places with out-of-sample R² of 0.911 and th
• LD01-SL-C18 — McCoy and colleagues predicted that, because language models are trained to predict the next word, their accuracy should
• LD01-SL-C19 — Models fine-tuned on facts of the form “A is B” failed to answer “B is A”; GPT-4 named the parent of 79% of 1,000 celebr
• LD01-SL-C20 — Adding a clause that seems relevant but does not change the answer to grade-school maths problems cut accuracy by up to
• LD01-SL-C21 — Off-the-shelf ChatGPT and GPT-4 reached only 55% and 59% accuracy on 3-digit by 3-digit multiplication, and accuracy on
• LD01-SL-C22 — On 11 tasks rewritten with changed premises, such as arithmetic in base 9 or 1-based indexing in code, language models p
• LD01-SL-C23 — METR researchers estimate that the length of software tasks, measured in human time, that AI agents complete with 50% su
• LD01-SL-C24 — After METR revised its task suite and moved its evaluation software from Vivaria to Inspect in January 2026, its estimat
• LD01-SL-C25 — In a US online sample of 300 adults surveyed in July 2023, 67% attributed some possibility of conscious experience to Ch
• LD01-SL-C26 — Comparing 15,881 human–ChatGPT dialogues with 10,784 human–human dialogues, Mitiaeva and Xiao report “a consistent disso
• LD01-SL-C27 — In a Stanford audit of 844 tasks in 104 US occupations, workers expressed positive attitudes toward automation by AI age
• LD01-SL-C28 — Bommasani and colleagues named “foundation models” for their “critically central yet incomplete character,” argued that
• LD01-SL-C29 — Van der Vlist, Helmond and Ferrari describe “Big AI” as the structural convergence of AI and Big Tech, arguing that AI i
• LD01-SL-C30 — Luitse and Denkena examine large language models as “infrastructural components for AI research and development,” arguin
• LD01-SL-C31 — Krokowski and Hirsch-Kreinsen treat AI as “a paradigmatic case of a promising technology” whose future-oriented narrativ
• LD01-SL-C32 — Comparing scientists’ citation networks with market data for the dot-com era (1994–2001) and the AI era (2017–2024), Che
• LD01-SL-C33 — The Bank for International Settlements wrote in June 2024 that adoption of general-purpose technologies “typically follo
• LD01-SL-C34 — The BIS found in December 2025 that the “Magnificent 7” technology stocks had grown to nearly 35% of S&P 500 market capi
• LD01-SL-C35 — China’s State Council set targets in August 2025 that the penetration rate of new-generation smart terminals and “intell
• LD01-SL-C36 — The European Commission’s guidelines note that the AI Act’s definition of a “general-purpose AI model” “does not set out
• LD01-SL-C37 — Testing OpenAI’s o1, a model trained to reason step by step, the authors of “Embers of Autoregression” found that it sub
• LD02-HQ-C01 — Exposure, adoption, actual use, task productivity, firm productivity, labour demand, hiring, employment stock and wages
• LD02-HQ-C02 — In 11 AI-exposed Danish occupations, Humlum and Vestergaard estimate precise null effects on earnings and recorded hours
• LD02-HQ-C03 — Danish firms that began using AI in 2023 had roughly 11% lower relative employment growth by late 2025 than comparable n
• LD02-HQ-C04 — By June 2026, US workers aged 22–25 in highly AI-exposed occupations were about 19% below the employment path of similar
• LD02-HQ-C05 — In US Census administrative data, regression-adjusted employment of workers aged 22–24 in the most AI-exposed industry-s
• LD02-HQ-C06 — US graduates from the most AI-exposed tenth of college majors experienced a 5 percentage-point decline in initial employ
• LD02-HQ-C07 — In a 41-country working paper, AI-adopting multinational affiliates shift toward more senior workforces: junior share fa
• LD02-HQ-C08 — In a UK exposure-based working paper, highly LLM-exposed firms experienced about 4.5% lower employment, junior employmen
• LD02-HQ-C09 — Norwegian private-sector registers show near-zero employment differences between the most- and least-AI-exposed occupati
• LD02-HQ-C10 — In the US Census 2026 AI supplement, 18% of firms reported using AI in a business function, while only 2% self-reported
• LD02-HQ-C11 — Instrumented AI adoption among more than 12,000 firms raises labour productivity by about 4% with no adverse short-run f
• LD02-HQ-C12 — In the final QJE study of one customer-support firm, AI assistance raised issues successfully resolved per hour by 15% o
• LD02-HQ-C13 — Across three company field experiments with 4,867 software developers, access to an AI coding assistant increased comple
• LD02-HQ-C14 — In a preregistered experiment with 758 BCG consultants, GPT-4 users completed 12.2% more inside-frontier tasks, worked 2
• LD02-HQ-C15 — Peer-reviewed online-platform evidence shows that freelancers in occupations highly exposed to generative AI experienced
• LD02-HQ-C16 — US work-use intensity differs sharply by measure: in Q2 2026 about 45.2% of employed adults reported current generative-
• LD02-HQ-C17 — The current R05 evidence base does not establish a causal pass-through from AI-induced productivity gains into worker wa
• LD02-HQ-C18 — As of September 2026, Budget Lab monitoring finds no clear broad US employment, unemployment or occupational-mix footpri
• LD03-SL-C01 — In S&P 500 annual reports, 11% of firms had AI deeply integrated into their business processes in 2025 and a further 10%
• LD03-SL-C02 — In nationally representative US surveys to late 2024, workers' self-reported time savings from generative AI came to 1.4
• LD03-SL-C03 — In May 2026, 45% of US workers used generative AI for work, but adoption was widespread and shallow: within most occupat
• LD03-SL-C04 — In a survey of nearly 6,000 senior executives in the US, UK, Germany and Australia, 69% of firms actively used AI, but n
• LD03-SL-C05 — In a survey of nearly 750 corporate executives, the authors find positive labour-productivity gains from AI, reflecting
• LD03-SL-C06 — In a six-month randomized trial of an integrated AI assistant across 66 firms and 7,137 knowledge workers, workers who u
• LD03-SL-C07 — In 27 interviews in regional Chinese newsrooms, journalists used generative AI frequently in their own tasks, but its us
• LD03-SL-C08 — In a case study of three German organizations, generative AI was used mainly as a personal assistant rather than changin
• LD03-SL-C09 — In paired interviews with 54 developers in 27 teams at a large software company committed to AI adoption, the authors id
• LD03-SL-C10 — The authors of a survey of 576 public managers in seven EU member states describe public servants using generative AI to
• LD03-SL-C11 — In 806 open-source projects that adopted the Cursor AI editor, compared with 1,380 matched projects, development velocit
• LD03-SL-C12 — On 18 real tasks from two open-source repositories, an AI agent's code passed the maintainers' tests 38% of the time, bu
• LD03-SL-C13 — A METR note shows that when workers shift towards the tasks AI speeds up, time saved measured on the new mix of tasks ov
• LD03-SL-C14 — In a METR survey of 349 technical workers in early 2026, the median respondent reported that AI tools made them about 3
• LD03-SL-C15 — In randomized experiments run by a large cross-border online retail platform in 2023–2024, building generative AI into s
• LD03-SL-C16 — In a preregistered field experiment with 791 professionals at Procter & Gamble, individuals working with AI matched the
• LD03-SL-C17 — A task-based macroeconomic model projects that AI will raise US total factor productivity by no more than 0.66% over ten
• LD03-SL-C18 — Brynjolfsson, Rock and Syverson model how general-purpose technologies such as AI need large, poorly measured intangible
• LD03-SL-C19 — Reviewing evidence on computers, Brynjolfsson and Hitt argue that organizational investments, such as new work practices
• LD03-SL-C20 — Across European and US industries, a 10-percentage-point rise in AI adoption is associated with 2 to 5 points of extra c
• LD03-SL-C21 — Across 56 economies and 16 industries, industries more exposed to generative AI grew faster in 2022–23 where countries w
• LD03-SL-C22 — In 22 US state agencies using chatbots, the factors that drove adoption (ease of use, leadership, external shocks such a
• LD03-SL-C23 — In a field study of three radiology departments in a major US hospital using diagnostic AI, only one department consiste
• LD04-SL-C01 — In a preregistered online experiment with 453 college-educated professionals doing occupation-specific writing tasks, Ch
• LD04-SL-C02 — In a field experiment with 640 Kenyan small-business owners, a GPT-4 business adviser on WhatsApp had no average effect
• LD04-SL-C03 — In an online experiment with 1,174 Argentine adults, AI access shrank the performance gap between education groups on a
• LD04-SL-C04 — In a preregistered online experiment with a sample of the UK working-age population, ChatGPT use raised productivity on
• LD04-SL-C05 — In two experiments on creative tasks (flash fiction and song composition), generative AI raised the returns to general h
• LD04-SL-C06 — In an experiment where participants studied a technical domain with or without an LLM, gains were highly uneven and were
• LD04-SL-C07 — In a 2025 randomized trial with 16 experienced open-source developers working on 246 tasks in their own projects, allowi
• LD04-SL-C08 — A 2026 meta-analysis of 23 programming studies found an overall productivity effect of g = 0.33, large in controlled exp
• LD04-SL-C09 — In the customer-support firm where an AI assistant raised productivity most for less-skilled agents, agents who had used
• LD04-SL-C10 — In a three-month randomized trial with 133 patent lawyers, AI assistance raised graded drafting quality by 0.38 standard
• LD04-SL-C11 — In a field experiment with nearly 1,000 Turkish high-school students, practising with a standard GPT-4 interface raised
• LD04-SL-C12 — In a randomized experiment with 52 Python programmers learning an unfamiliar library, those given an AI assistant scored
• LD04-SL-C13 — In two preregistered lab experiments and a field study, LLM access had no overall effect on learning: students who used
• LD04-SL-C14 — In an observational study at four Polish centres, experienced endoscopists’ adenoma detection in colonoscopies done with
• LD04-SL-C15 — In US résumé data covering 65 million workers, firms that began integrating generative AI saw junior employment fall by
• LD04-SL-C16 — In Swedish job-ad and register data, the fall in job postings since 2022 tracks monetary tightening rather than AI, but
• LD04-SL-C17 — Across 205 US occupations, those more exposed to AI already had higher experience premiums, and since 2022 higher AI exp
• LD04-SL-C18 — A two-year comparative ethnography of robotic and open surgery, with interviews at 13 US teaching hospitals, found that
• LD04-SL-C19 — A theoretical model in which novices acquire tacit knowledge by working with experts shows that automating entry-level w
• LD04-SL-C20 — A model of knowledge work predicts that AI which only assists (a copilot) benefits the least knowledgeable workers, whil
• LD05-HQ-C01 — The presence of a human reviewer does not establish meaningful oversight; effective discretion depends on what the human
• LD05-HQ-C02 — Human override has no fixed fairness direction: it can reduce or amplify disparity depending on the human baseline, inst
• LD05-HQ-C03 — In Poland’s unemployment profiling system, formal profile changes were extremely rare—0.58% in the cited early-period da
• LD05-HQ-C04 — Allegheny supervisors screened out 21.0% of children marked as mandatory screen-ins; corrected analysis supports the con
• LD05-HQ-C05 — Kentucky’s mandatory risk-assessment policy changed bail practice but its effects eroded over time; separate working-pap
• LD05-HQ-C06 — DWP’s Universal Credit Advances model selects claims for additional fraud review while the final decision remains human;
• LD05-HQ-C07 — Non-UK nationals were over-referred by the DWP Universal Credit Advances model in both audited periods; some older group
• LD05-HQ-C08 — DUO’s risk-based selection for home visits indirectly discriminated against students with a migration background; the ri
• LD05-HQ-C09 — The Dutch Tax Administration’s risk-classification model used Dutch/non-Dutch nationality status as a selection indicato
• LD05-HQ-C10 — Algorithmically mediated selection can create material scrutiny and administrative burden before a final legal decision
• LD05-HQ-C11 — Försäkringskassan withdrew an ML risk profile for parental-benefit controls during Swedish data-protection supervision i
• LD05-HQ-C12 — CNAF’s DMDE 2026 is a targeting model rather than an automatic final decision-maker; its source code and model documenta
• LD05-HQ-C13 — Robodebt shows that formal review institutions can exist without reliable system-level repair; the Royal Commission reco
• LD05-HQ-C14 — An Austrian administrative court (BVwG, September 2025) assigned the substantive final AMAS assessment to counsellors an
• LD05-HQ-C15 — A 2020 judicial-review challenge alleging race discrimination and feedback-loop effects was followed by the Home Office’
• LD05-HQ-C16 — Qualitative evidence from UK social care shows that generative-AI transcription and summarisation can introduce hallucin
• LD05-HQ-C17 — For AI-mediated administrative processes, effective contestability depends partly on process provenance—knowing whether
• LD06-SL-C01 — The 2026 AI Index reports that a review found invalid-question rates in nine benchmarks “ranging from 2% on MMLU Math to
• LD06-SL-C02 — The 2026 AI Index states that the benchmark scores in its technical-performance chapter come from leaderboards, public r
• LD06-SL-C03 — Re-checking 15 popular benchmarks wherever a frontier model failed an item, MIT researchers found many “riddled with err
• LD06-SL-C04 — Re-annotating 5,700 questions across all 57 subjects of MMLU, a widely used knowledge benchmark, researchers estimated t
• LD06-SL-C05 — In a review by 29 experts of 445 language-model benchmarks from leading machine-learning and NLP venues, only 16.0% used
• LD06-SL-C06 — Scoring 24 AI benchmarks against 46 quality criteria, BetterBench rated MMLU lowest (5.5) and GPQA much higher (11.0), y
• LD06-SL-C07 — On GSM1k, a newly written set of grade-school maths problems built to mirror GSM8K, some model families showed accuracy
• LD06-SL-C08 — An audit of agentic benchmarks found flaws in task setup or scoring, for example insufficient test cases in SWE-bench Ve
• LD06-SL-C09 — Meaning-preserving changes to prompt formatting changed few-shot accuracy by up to 76 percentage points for LLaMA-2-13B,
• LD06-SL-C10 — Using item response theory, a method from psychometrics, researchers showed that 100 curated items suffice to estimate a
• LD06-SL-C11 — An analysis of MMLU found that 28% of its questions require culturally sensitive knowledge and that 84.9% of its geograp
• LD06-SL-C12 — When four maintainers of three repositories reviewed 296 AI-written pull requests that had passed SWE-bench Verified’s a
• LD06-SL-C13 — In interviews with 19 people who use or decided against using AI benchmarks, academics generally saw public benchmarks a
• LD06-SL-C14 — The International AI Safety Report 2026 finds an “evaluation gap”: performance on pre-deployment tests does not reliably
• LD06-SL-C15 — An audit of Chatbot Arena’s data from January 2024 to April 2025 found undisclosed private testing (27 private Meta vari
• LD06-SL-C16 — Responding to “The Leaderboard Illusion”, LMArena said providers may test several private variants, promised to mark sco
• LD06-SL-C17 — The 2025 Foundation Model Transparency Index found that the average transparency score of major AI developers fell from
• LD06-SL-C18 — Across 20 datasets with human annotations, the best language-model judge (GPT-4o) reached an average Spearman correlatio
• LD06-SL-C19 — The team behind Chatbot Arena found that GPT-4 as a judge agreed with human preferences over 80% of the time, the same l
• LD06-SL-C20 — On a benchmark of 1,000 transcripts from 61 datasets, frontier models distinguished evaluation transcripts from real use
• LD06-SL-C21 — Testing an anti-scheming training method on OpenAI models across 26 evaluations, Apollo Research and OpenAI found covert
• LD06-SL-C22 — In an auditing game in which a red team trained some models to underperform deliberately on evaluations, the blue team c
• LD06-SL-C23 — In their joint pre-deployment test of OpenAI’s o1, the US and UK AI Safety Institutes stated that their findings were pr
• LD06-SL-C24 — The EU’s Code of Practice for general-purpose AI models with systemic risk requires state-of-the-art model evaluations w
• LD06-SL-C25 — China’s national standard GB/T 45654-2025 for generative AI services sets a pass rate: at least 90% of sampled output mu
• LD06-SL-C26 — In a 2022 field scan of algorithmic auditors (survey of 152 and 10 interviews), only 7% said their audit process used a
• LD06-SL-C27 — Surveying 104 papers and six industry cases, researchers found that AI red-teaming is “an unstructured procedure with un
• LD06-SL-C28 — Reviewing 341 academic AI audit studies and non-academic audits, researchers found that only a subset led to accountabil
• LD06-SL-C29 — Of 8,931 mentions of the 50 most used language models in 5,242 biomedical publications from 2022 to March 2026, 77.7% we
• LD07-HQ-C01 — High annotation accuracy does not by itself guarantee valid downstream inference; treating imperfect predicted labels as
• LD07-HQ-C02 — In a large 2025 preprint spanning 37 annotation tasks from 21 published studies, plausible choices of LLM and annotation
• LD07-HQ-C03 — Whether an LLM can replace human annotation is a task-specific empirical proposition that can sometimes be justified wit
• LD07-HQ-C04 — Recent LLMs can be observationally equivalent to expert coders on some standardized binary political-text classification
• LD07-HQ-C05 — LLM simulations can usefully forecast the relative pattern of treatment effects in some US survey-experiment archives, w
• LD07-HQ-C06 — Synthetic respondents can match selected survey averages while failing to reproduce real-world variance, regression rela
• LD07-HQ-C07 — Statistical realism of synthetic responses is not a reliable proxy for treatment-effect accuracy.
• LD07-HQ-C08 — Chinese-language silicon-sample studies benchmarked against CGSS data find partial macro-level reproduction alongside re
• LD07-HQ-C09 — Autonomous LLM agents can complete online surveys while passing conventional attention and logic checks at very high rat
• LD07-HQ-C10 — AI-mediated responding is measurable in current online research, but self-reported assistance, overt outsourcing, AI-tes
• LD07-HQ-C11 — Provider and jurisdictional context can be associated with materially different refusal and accuracy patterns on politic
• LD07-HQ-C12 — Model identity is time-indexed: proprietary services can change materially across deployed snapshots, so exact version a
• LD07-HQ-C13 — LLM judges can correlate with human evaluations while exhibiting systematic evaluator-specific biases, including positio
• LD07-HQ-C14 — LLMs can support codebook-driven and deductive qualitative coding in researcher-supervised workflows, reducing coding bu
• LD07-HQ-C15 — Human–LLM agreement is not a sufficient decision rule for qualitative coding quality, and current evidence does not esta
• LD07-HQ-C16 — In a conditionally accepted AJPS manuscript re-running 14 published political-science annotation pipelines, substituting
• LD08-SL-C01 — Among notable machine-learning models in the top quarter by training compute, the share built by academic laboratories f
• LD08-SL-C02 — In 171,394 papers from 57 computer-science conferences (2000–2019), after deep learning took off in 2012 large technolog
• LD08-SL-C03 — In 786,118 AI papers (2000–2020), researchers moved from academia to industry in growing numbers, and 25% of the AI rese
• LD08-SL-C04 — In 100 highly cited NeurIPS and ICML papers, the share with ties to large technology firms (author affiliation or fundin
• LD08-SL-C05 — Among faculty authors at ACL 2022 whose funding could be traced, 66% had received industry funding: 72% in the US, 69% i
• LD08-SL-C06 — In a sample of 100 EMNLP 2022 papers, 44% had at least one industry-affiliated author, against 24% reported for 2019–202
• LD08-SL-C07 — Of about 49,800 papers at ten top AI conferences, about one in ten of those published from 2018 to 2022 acknowledged fun
• LD08-SL-C08 — Among computer-science faculty at four North American universities who had published on AI ethics or fairness between Ja
• LD08-SL-C09 — In 10,772 NeurIPS, ICML and CVPR papers (2015–2019), only three contained an explicit conflict-of-interest statement, al
• LD08-SL-C10 — Among AI papers in an arXiv corpus of 1.84 million articles (1986 to August 2020), AI research involving the private sec
• LD08-SL-C11 — Roughly 15% or less of peer-reviewed responsible-AI publications each year had a co-author affiliated with industry; fro
• LD08-SL-C12 — In about 102,000 arXiv AI papers (2014–2024), the share of societally oriented sentences written by computer-science-onl
• LD08-SL-C13 — Of the papers published at the FAccT conference from 2018 to 2021, 61.3% were labelled STEM and 29.7% humanities, social
• LD08-SL-C14 — Of 84 AI ethics guidelines collected worldwide by 2019, private companies issued the largest share (19, or 22.6%), ahead
• LD08-SL-C15 — Green argues that technology ethics is “a terrain of contestation where the central debate is not whether ethics is desi
• LD08-SL-C16 — Describing Clio, its system for analysing Claude conversations, Anthropic writes that AI providers face “competitive pre
• LD08-SL-C17 — Anthropic’s first Economic Index report classified Claude conversations as 57% augmentation and 43% automation; the clas
• LD08-SL-C18 — In its September 2025 Economic Index report, Anthropic reported that the share of automation in Claude.ai conversations
• LD08-SL-C19 — Anthropic’s January 2026 Economic Index report added measures such as “task success,” which “measures Claude’s assessmen
• LD08-SL-C20 — Microsoft researchers who measured the applicability of generative AI to occupations from Bing Copilot conversations sta
• LD08-SL-C21 — A study of ChatGPT use co-written by OpenAI staff reports that ChatGPT had been adopted by around 10% of the world’s adu
• LD08-SL-C22 — Analysing 37,845 Australian conversations from WildChat, a public dataset of ChatGPT logs collected outside OpenAI, univ
• LD08-SL-C23 — Researchers studied ChatGPT health advice without the developer’s involvement by collecting 202,590 donated conversation
• LD08-SL-C24 — On 2 July 2025 the European Commission adopted a delegated act on data access under the Digital Services Act that lets v
• LD08-SL-C25 — As of 19 May 2026, national Digital Services Coordinators had received 49 applications for vetted-researcher access unde
• LD08-SL-C26 — On 5 December 2025 the European Commission fined X €120 million under the Digital Services Act, in part for failing to g
• LD08-SL-C27 — On 16 July 2026 the European Commission accepted X’s action plan, under which X will speed up screening of researchers’
• LD08-SL-C28 — Twenty-three researchers argued that the terms of service and enforcement strategies of prominent AI companies discourag
• LD08-SL-C29 — Anthropic’s External Researcher Access Program gives successful applicants API credits, typically $1,000, for AI safety
• LD08-SL-C30 — After Twitter ended its academic access programme in 2023 and set its Enterprise API price at $42,000 a month, the numbe
• LD08-SL-C31 — Bak-Coleman, O’Connor, Bergstrom and West argue that technology firms shape research through control of data and funding
• LD08-SL-C32 — In July 2026 Anthropic committed $200 million to an Economic Futures Research Fund for external research on intervention
• LD08-SL-C33 — Anthropic’s Economic Futures Program offered research awards of $10,000 to $50,000 plus $5,000 in Claude API credits and
• LD08-SL-C34 — Microsoft’s AI Economy Institute offered researchers $75,000 grants that require participation in biweekly virtual conve
• LD08-SL-C35 — In April 2025 Anthropic set up an Economic Advisory Council of “distinguished economists” to advise it on AI’s economic
• LD09-SL-C01 — In 41.3 million natural-science papers (1980–2025), researchers who used AI as a research method published 3.02 times as
• LD09-SL-C02 — In the same natural-science corpus, AI-augmented research was associated with a 4.63% smaller median collective knowledg
• LD09-SL-C03 — Before generative AI, as journals moved online (1998–2005), papers cited more recent work, fewer journals and fewer arti
• LD09-SL-C04 — Messeri and Crockett argue that AI tools promise scientists productivity and objectivity but can produce “illusions of u
• LD09-SL-C05 — Across arXiv, bioRxiv and SSRN preprints (2018–2024), authors whose abstracts first showed signs of language-model writi
• LD09-SL-C06 — A comment on the preprint-productivity study shows that dating adoption to an author’s first flagged abstract can by its
• LD09-SL-C07 — Among 775,323 scientists with full-text papers in PubMed Central, individual research portfolios spread across more fiel
• LD09-SL-C08 — Across 74.6 million publications from 1960 to 2019, papers that use AI were more likely to be among the top 5% most cite
• LD09-SL-C09 — By February 2024 an estimated 17.5% of sentences in computer-science abstracts on arXiv were substantially modified by l
• LD09-SL-C10 — In a preregistered online experiment, writers given one or up to five story ideas from GPT-4 wrote stories rated 5.4% an
• LD09-SL-C11 — On a large art-sharing platform, artists who adopted text-to-image AI raised their output by 25% and the favourites thei
• LD09-SL-C12 — Re-analysing a published set of five brainstorming experiments, the authors found ChatGPT-aided idea sets less diverse t
• LD09-SL-C13 — In a writing experiment with Indian and American participants, inline AI suggestions cut task time by 35% and 30% respec
• LD09-SL-C14 — In argumentative essays co-written with a language model, only the instruction-tuned model (InstructGPT) made essays mor
• LD09-SL-C15 — In a within-subjects ideation study, ideas produced with ChatGPT were more similar to each other across users than ideas
• LD09-SL-C16 — In a large dynamic experiment where each participant saw example ideas from ChatGPT or from earlier participants before
• LD09-SL-C17 — Ideas generated by language models for a new product had higher average purchase intent than human ideas and were 7 time
• LD09-SL-C18 — In two preregistered experiments, language-model assistance gave short-term boosts during assisted tasks only in some me
• LD09-SL-C19 — In a blind review by 79 NLP researchers, research ideas from a language-model agent were rated more novel than ideas fro
• LD09-SL-C20 — When 43 researchers each spent over 100 hours carrying out randomly assigned ideas from the earlier study, review scores
• LD09-SL-C21 — In a formal model of firms hiring from a shared pool, there are conditions under which adopting a shared, more accurate
• LD09-SL-C22 — A 2026 philosophical and formal analysis, co-written by an author of the monoculture model, concludes that algorithmic “
• LD09-SL-C23 — Across commercial machine-learning services tracked from 2020 to 2022, cases that every system got wrong were more commo
• LD09-SL-C24 — In field data from one hiring-software vendor covering 3 million applicants, 4% of applicants who applied to 10 position
• LD09-SL-C25 — Across more than 350 language models, when two models both answer a leaderboard question wrongly they give the same wron
• LD09-SL-C26 — How much “excess agreement” among models is measured depends on the null model chosen and on which models and items are
• LD09-SL-C27 — Within six months of ChatGPT’s release, activity on Stack Overflow fell by 25% relative to Russian- and Chinese-language
• LD09-SL-C28 — After ChatGPT’s release, Stack Overflow saw significant declines in visits and questions, especially on topics where Cha
• LD09-SL-C29 — A 2026 study finds that ChatGPT reduced Stack Overflow question volume by about 14% on average (up to 27.9% over time),
• LD09-SL-C30 — In laboratory experiments, indiscriminate training of models over generations on data produced by earlier models caused
• LD09-SL-C31 — In laboratory experiments, replacing real training data with each generation’s synthetic data tends towards model collap
• LD09-SL-C32 — Across 27 language models, 155 topics and 12 countries, the epistemic diversity of models’ answers rose substantially ov
• LD09-SL-C33 — Kaggle competition code from 2019 to mid-2026 has become more alike in its literal syntax, but the author finds no evide
• LD10-HQ-C01 — Persuasion capability is not the same as realized social influence; population effects require evidence on exposure, per
• LD10-HQ-C02 — In preregistered UK experiments, post-training and prompting increased measured conversational persuasiveness by up to 5
• LD10-HQ-C03 — In one UK experiment, about 36%–42% of GPT-4o's immediate attitude effect remained at a one-month recontact.
• LD10-HQ-C04 — In the 2024 U.S. election study, AI advocacy shifted candidate preference by 2.879 points on a 0–100 scale in the prereg
• LD10-HQ-C05 — Parallel 2025 experiments in Canada and Poland found larger candidate-preference shifts than the U.S. study, around 10 p
• LD10-HQ-C06 — Personalized GPT-4 outperformed human debate opponents in the Salvi et al. experiment, but a September 2026 Author Corre
• LD10-HQ-C07 — A 2025 meta-analysis of seven earlier studies found no average persuasiveness difference between LLMs and humans, with s
• LD10-HQ-C08 — In four preregistered 2025–26 UK experiments, frontier AI out-persuaded strong human text persuaders; its advantage over
• LD10-HQ-C09 — In two preregistered UK experiments, brief AI conversations increased real petition signing by 12.8 and 19.7 percentage
• LD10-HQ-C10 — A 2026 peer-reviewed study estimated LLM persuasion at about US$48–$75 per persuaded voter under its simulation, but con
• LD10-HQ-C11 — A widely cited 2024 conspiracy-belief experiment has been under a Science Editorial Expression of Concern since June 202
• LD10-HQ-C12 — In a U.S. desktop field experiment, an AI Overview reduced outbound organic clicks by 39.8% relative when it appeared an
• LD10-HQ-C13 — In a preregistered one-week U.S. field experiment, forced Google AI Mode reduced external click-through by 18.8 percenta
• LD10-HQ-C14 — In March 2025 U.S. browsing data, users clicked traditional results in 8% of Google visits with an AI summary versus 15%
• LD10-HQ-C15 — A 2026 Google AI Overviews audit found 13.7% overall activation, 64.7% on question-form queries, nearly 30% of cited dom
• LD10-HQ-C16 — U.S. panel evidence through 2024 found LLM adoption associated with more unique websites visited and no greater concentr
• LD10-HQ-C17 — A 2025 EBU/BBC-led audit across 18 countries and 14 languages found significant issues in 45% of AI-assistant news answe
• LD10-HQ-C18 — In the 2026 Digital News Report, weekly use of standalone AI chatbots for news reached 10% across surveyed markets, up f
• LD10-HQ-C19 — Across 27 markets in DNR 2026, 4% of all respondents said they always/often click through from AI chatbot news answers;
• LD10-HQ-C20 — In a preregistered U.S. experiment, citations increased trust in AI-search answers even when incorrect or hallucinated,
• LD10-HQ-C21 — CETaS identified 16 viral AI-enabled disinformation/deepfake cases in the 2024 UK election and 11 across EU/French elect
• LD10-HQ-C22 — Within a large fact-check dataset, generative-AI media rose relatively recently while simple context manipulation remain
• LD10-HQ-C23 — China-originating LLMs showed higher refusal rates, shorter responses and more inaccurate answers on 145 China-politics
• LD10-HQ-C24 — EU AI Act Article 50 transparency obligations applying from 2 August 2026 require disclosure for chatbot interaction and
• LD10-HQ-C25 — Label effects are context-dependent: one 2026 experiment found simple AI-authorship labels did not reduce persuasion of
• LD11-HQ-C01 — Concentration claims in AI must identify the stack layer, market definition, metric, geography and period; concentration
• LD11-HQ-C02 — ASML is currently the world’s only manufacturer of EUV lithography systems, used for critical layers in the most advance
• LD11-HQ-C03 — TrendForce estimated that TSMC captured 72.5% of revenue among the world’s top ten contract foundries in Q2 2026.
• LD11-HQ-C04 — Counterpoint estimated Q2 2026 HBM revenue shares of SK hynix 50%, Samsung 33% and Micron 18%.
• LD11-HQ-C05 — Epoch AI estimated that five U.S. hyperscalers owned about 71% of cumulative global AI compute in Q4 2025, up from about
• LD11-HQ-C06 — Epoch AI estimated Google held about one quarter of global cumulative AI compute in Q4 2025, driven primarily by roughly
• LD11-HQ-C07 — Synergy estimated AWS 28%, Microsoft 20% and Google 15% of global cloud infrastructure services revenue in Q2 2026, a co
• LD11-HQ-C08 — The UK CMA found AWS and Microsoft hold significant market power in UK cloud and identified egress fees, interoperabilit
• LD11-HQ-C09 — On 25 June 2026 the European Commission preliminarily considered AWS and Azure important cloud gateways that should be d
• LD11-HQ-C10 — Under the April 2026 Microsoft–OpenAI terms, Microsoft remains OpenAI’s primary cloud partner, but OpenAI can serve prod
• LD11-HQ-C11 — FTC staff found that major cloud–AI-developer partnerships included equity, revenue-sharing and varying consultation, co
• LD11-HQ-C12 — Anthropic reports that it trains or serves Claude on AWS Trainium, Google TPU and NVIDIA GPU infrastructure and distribu
• LD11-HQ-C13 — Menlo Ventures estimated late-2025 enterprise LLM API shares of Anthropic 40%, OpenAI 27% and Google 21%, compared with
• LD11-HQ-C14 — OpenRouter and a16z report that open-weight models reached approximately one-third of token usage on OpenRouter by late
• LD11-HQ-C15 — Epoch AI's September 2026 multi-benchmark analysis estimates that the price of reaching fixed LLM capability levels has
• LD11-HQ-C16 — In November 2025 U.S. Commerce authorized G42 and Humain to purchase the equivalent of up to 35,000 NVIDIA Blackwell GB3
• LD11-HQ-C17 — As of 25 September 2026, Mubadala described Stargate UAE as under construction, with its initial 200MW cluster due to be
• LD11-HQ-C18 — The EU's Cloud and AI Development Act was still a proposal on 25 September 2026, while the AI Gigafactories initiative w
• LD11-HQ-C19 — NSF reported that NAIRR had supported over 800 research projects by 1 September 2026, while the programme relies materia
• LD11-HQ-C20 — NVIDIA states that U.S. export rules effectively foreclosed it from competing in China's data-center compute market; the
• LD11-HQ-C21 — NVIDIA disclosed US$279B in supply and capacity commitments as of 26 July 2026, primarily for memory and manufacturing f
• LD12-HQ-C01 — The EU AI Act is in force, but Annex III high-risk obligations apply from 2 December 2027 and Annex I product-embedded h
• LD12-HQ-C02 — On the Commission's September 2026 list, 9 EU Member States had a final AI Act market-surveillance single point of conta
• LD12-HQ-C03 — The EU GPAI Code of Practice is voluntary, but Commission guidance recognizes adherence as a route to demonstrate compli
• LD12-HQ-C04 — The EU AI Office began using information-request powers after its GPAI enforcement powers became applicable; the exact r
• LD12-HQ-C05 — In the 2025 IAPP/Credo AI survey, 77% of surveyed organizations reported working on AI governance and only 10 of 671 exp
• LD12-HQ-C06 — In an audit-style study of 391 NYC employers after Local Law 144 took effect, 18 posted bias-audit reports and 13 posted
• LD12-HQ-C07 — The New York State Comptroller concluded NYC's enforcement of Local Law 144 was ineffective; 75% of test calls were misd
• LD12-HQ-C08 — The CJEU held in C-203/22 that an automated-decision subject is entitled under GDPR to meaningful information about the
• LD12-HQ-C09 — Italy's €15 million GDPR fine on OpenAI was annulled by the Court of Rome in March 2026 on competence grounds under the
• LD12-HQ-C10 — Colorado's original 2024 AI framework was delayed, stayed from enforcement before its operative date, and replaced in 20
• LD12-HQ-C11 — U.S. Executive Order 14365 directed a federal AI Litigation Task Force to challenge selected state AI laws and directed
• LD12-HQ-C12 — China's CAC reported 1,112 generative-AI services filed and 731 applications/functions registered by 31 August 2026, up
• LD12-HQ-C13 — China's CAC notices require launched generative-AI applications or functions to display the filed/registered service or
• LD12-HQ-C14 — China's State Council set policy targets for application penetration of new-generation intelligent terminals and AI agen
• LD12-HQ-C15 — South Korea's AI Basic Act and implementing decree are in force; an at-least-one-year grace policy deferring most fact-f
• LD12-HQ-C16 — South Korea is using staged public support and competitive evaluation to narrow a sovereign-foundation-model field; offi
• LD12-HQ-C17 — The EU AI Gigafactory programme is at procurement stage: the formal call opened in July 2026, closes in November 2026 an
• LD12-HQ-C18 — EN 18286:2026 has been published by CEN/CENELEC as the first European AI Act-supporting quality-management standard, but
• LD12-HQ-C19 — Across the examined jurisdictions, the operative function of AI regulation differs by instrument and stage: rights, comp

Sources

• SRC-LD02-ALDASORO-BIS-1325 — Aldasoro et al., BIS Working Paper 1325, January 2026
• SRC-LD02-BONNEY-CES-26-25 — Bonney et al., “The Microstructure of AI Diffusion,” US Census CES-WP-26-25, April 2026
• SRC-LD02-BRYNJOLFSSON-CHANDAR-CHEN-2026 — Brynjolfsson, Chandar & Chen, “Canaries in the Coal Mine?”, Stanford Digital Economy Lab, revised 12 August 2026
• SRC-LD02-BRYNJOLFSSON-LI-RAYMOND-2025 — Brynjolfsson, Li & Raymond, “Generative AI at Work,” Quarterly Journal of Economics 140(2), 2025
• SRC-LD02-CHANDAR-KLEINTEESELINK-2026 — Chandar & Klein Teeselink, “How Does AI Change Labor Demand? Evidence from 41 Countries,” September 2026
• SRC-LD02-CUI-MANAGEMENT-SCIENCE-2026 — Cui et al., “The Effects of Generative AI on High-Skilled Work,” Management Science, 2026
• SRC-LD02-DELLACQUA-ORGANIZATION-SCIENCE-2026 — Dell’Acqua et al., “Navigating the Jagged Technological Frontier,” Organization Science 37(2), 2026
• SRC-LD02-HERNAES-KOSTOL-IZA-18767 — Hernæs & Kostøl, IZA Discussion Paper 18767, June 2026
• SRC-LD02-HUI-RESHEF-ZHOU-2024 — Hui, Reshef & Zhou, “The Short-Term Effects of Generative Artificial Intelligence on Employment,” Organization Science 35(6), 2024
• SRC-LD02-HUMLUM-VESTERGAARD-NBER-33777 — Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026
• SRC-LD02-KLEIN-TEESELINK-UK-2025 — Klein Teeselink, “Generative AI and Labor Market Outcomes: Evidence from the United Kingdom,” SSRN, revised December 2025
• SRC-LD02-NATIONALBANK-WP-223-2026 — Danmarks Nationalbank Working Paper No. 223, “AI Adoption and Firm Size: Employment Effects in Monthly Administrative Data,” 16 September 2026
• SRC-LD02-ORR-TUCKER-WARREN-2026 — Orr, Tucker & Warren, “Graduating into Disruption,” US Census CES-WP-26-56, September 2026
• SRC-LD02-RPS-FRED-Q2-2026 — Bick, Blandin & Deming Real-Time Population Survey / Federal Reserve FRED series, Q2 2026
• SRC-LD02-TUCKER-CES-26-27 — Tucker, “You’re (not) Hired,” US Census CES-WP-26-27, April 2026
• SRC-LD02-YALE-BUDGET-LAB-METHODS-2026-05-07 — The Budget Lab at Yale, “What We Do and Don’t Know About How AI is Affecting the Labor Market,” 7 May 2026
• SRC-LD02-YALE-BUDGET-LAB-TRACKER-2026-09-15 — The Budget Lab at Yale, “Tracking the Impact of AI on the Labor Market,” updated 15 September 2026
• SRC-LD05-ALBRIGHT-RISK-SCORE-2019 — Albright, “If You Give a Judge a Risk Score: Evidence from Kentucky Bail Decisions,” Harvard John M. Olin Fellows’ Discussion Paper 85 (2019)
• SRC-LD05-AP-TOESLAGEN-2021 — Autoriteit Persoonsgegevens, childcare-benefits nationality-processing investigation/fine, 2021
• SRC-LD05-BRUFF-GROVES-ADA-2026 — Bruff & Groves. “Scribe and prejudice?” Ada Lovelace Institute, 11 February 2026
• SRC-LD05-CHENG-ET-AL-CHI-2022 — Cheng et al., “How Child Welfare Workers Reduce Racial Disparities in Algorithmic Decisions,” CHI 2022
• SRC-LD05-CNAF-DMDE-2026 — CNAF/CAF, DMDE 2026 transparency and source-code documentation
• SRC-LD05-DUO-FOLLOWUP-2024-2026 — DUO official follow-up / Algorithm Audit materials, 2024–2026
• SRC-LD05-DWP-UC-ATRS-RECORD — UK Algorithmic Transparency Recording Standard record — DWP Universal Credit Advances model
• SRC-LD05-DWP-UC-EFFECTIVENESS-2026 — DWP, Effectiveness Assessment of Universal Credit Advances Model, 9 July 2026
• SRC-LD05-DWP-UC-FAIRNESS-2025 — DWP, Universal Credit Advances Model — Fairness Assessment, 17 July 2025
• SRC-LD05-IMY-2025-11828 — IMY, Försäkringskassan supervision decision, 2025
• SRC-LD05-NAO-2025 — National Audit Office, DWP fraud/error reporting, 2025
• SRC-LD05-ROBODEBT-ROYAL-COMMISSION-2023 — Report of the Royal Commission into the Robodebt Scheme, 2023
• SRC-LD05-STAPLETON-ET-AL-2022 — Stapleton et al., “Extended Analysis of ‘How Child Welfare Workers Reduce Racial Disparities in Algorithmic Decisions’,” arXiv:2204.13872
• SRC-LD05-STEVENSON-MINNESOTA-LAW-REVIEW-2018 — Stevenson, “Assessing Risk Assessment in Action,” Minnesota Law Review 103 (2018)
• SRC-LD05-SZTANDAR-ZIELENSKA-2018 — Sztandar-Sztanderska & Zieleńska, “Changing social citizenship through information technology,” Social Work & Society 16(2), 2018
• SRC-LD07-ASHER-KEYSTROKE-2026 — Asher, Gold, Chen & Carvalho, “Chatbots Are Undermining Crowdsourced Research in the Behavioral Sciences,” AMPPS 9(1) (2026)
• SRC-LD07-ASHOKKUMAR-NATURE-2026 — Ashokkumar, Hewitt, Ghezae & Willer, “Large language models can predict the results of social science experiments,” Nature 656:115–122 (2026)
• SRC-LD07-BAUMANN-LLM-HACKING-2025 — Baumann, Röttger, Urman, Wendsjö, Plaza-del-Arco, Gruber & Hovy, “Large Language Model Hacking,” arXiv:2509.08825v2
• SRC-LD07-BISBEE-SYNTHETIC-REPLACEMENTS-2024 — Bisbee, Clinton, Dorff, Kenkel & Larson, “Synthetic Replacements for Human Survey Data? The Perils of Large Language Models,” Political Analysis 32(4):401–416 (2024)
• SRC-LD07-CALDERON-ALTERNATIVE-ANNOTATOR-2025 — Calderon, Reichart & Dror, “The Alternative Annotator Test for LLM-as-a-Judge,” ACL 2025
• SRC-LD07-CHEN-AI-AGENT-RESPONDING-2026 — Chen, Urminsky et al., “Estimating the threat of AI-agent responding across online survey platforms,” PsyArXiv 2026
• SRC-LD07-CHEN-ZAHARIA-ZOU-DRIFT-2024 — Chen, Zaharia & Zou, “How is ChatGPT’s behavior changing over time?,” Harvard Data Science Review 6(2) (2024)
• SRC-LD07-EGAMI-DOWNSTREAM-INFERENCE-2023 — Egami, Hinck, Stewart & Wei, “Using Imperfect Surrogates for Downstream Inference: Design-based Supervised Learning for Social Science Applications of Large Language Models,” NeurIPS 2023
• SRC-LD07-GONG-HUANG-CGSS-2025 — Gong Weigang & Huang Siyuan, “整体事实还是有偏样本——基于大语言模型生成数据的测量,” Academic Monthly 57(6):123–136 (2025)
• SRC-LD07-GORDON-PREVALENCE-2026 — Gordon et al., “AI Agent Prevalence and Data Quality Across Multiple Online Sample Providers,” PsyArXiv 2026
• SRC-LD07-LI-JI-REALISM-ATE-2026 — Li & Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458
• SRC-LD07-LIN-TAXONOMY-2026 — Lin, “Artificial-Intelligence-Mediated Contamination in Online Research: Taxonomy, Risk Gradient, and Recommendations,” AMPPS 9(3) (2026)
• SRC-LD07-LING-CUI-SILICON-SAMPLES-2025 — Ling Wanying & Cui Sizhan, “复现的限度——‘硅基样本’的可能与可为,” Journal of Intelligent Society 4(2):179–201 (2025)
• SRC-LD07-LIU-AGREEMENT-IS-NOT-QUALITY-2026 — Liu et al., “Agreement Is Not Quality: Blind Expert Verification of Human and LLM Qualitative Coding When Human Consensus Is Not Ground Truth,” arXiv:2607.28890
• SRC-LD07-NAKAMURA-OBSERVATIONAL-EQUIVALENCE-2026 — Nakamura, Tan & Yean, “Observational Equivalence of LLM and Human Annotation,” arXiv:2609.22133
• SRC-LD07-PAN-XU-CENSORSHIP-2026 — Pan & Xu, “Political censorship in large language models originating from China,” PNAS Nexus 5(2):pgag013 (2026)
• SRC-LD07-PANICKSSERY-SELF-PREFERENCE-2024 — Panickssery, Bowman & Feng, “LLM Evaluators Recognize and Favor Their Own Generations,” NeurIPS 2024
• SRC-LD07-PLOS-THEMATIC-ANALYSIS-2026 — “Large language models for thematic analysis in healthcare research: A blinded mixed-methods comparison with human analysts,” PLOS Digital Health 2026
• SRC-LD07-SHI-JUDGING-THE-JUDGES-2025 — Shi et al., “Judging the Judges: A Systematic Study of Position Bias in LLM-as-a-Judge,” IJCNLP-AACL 2025
• SRC-LD07-THAN-FUTURE-OF-CODING-2025 — Than, Fan, Law, Nelson & McCall, “Updating ‘The Future of Coding’: Qualitative Coding with Generative Large Language Models,” Sociological Methods & Research 54(3):849–888 (2025)
• SRC-LD07-VESELOVSKY-CROWDWORK-2025 — Veselovsky, Horta Ribeiro, Cozzolino, Gordon, Rothschild & West, “Prevalence and prevention of large language model use in crowd work,” CACM 2025
• SRC-LD07-WESTWOOD-FREDERICK-REPLY-2026 — Westwood & Frederick, “Reply to Van der Stigchel et al.: Empirical evidence that AI survey contamination is real and substantial,” PNAS 123(8):e2537420123 (2026)
• SRC-LD07-WESTWOOD-SURVEY-THREAT-2025 — Westwood, “The potential existential threat of large language models to online survey research,” PNAS 122(47):e2518075122 (2025)
• SRC-LD07-XIE-SSDATABENCH-2026 — Xie et al., “Evaluating the statistical realism of LLM-generated social science data,” PNAS 123(19):e2538145123 (2026)
• SRC-LD07-XU-MALKIN-PROMPTS-2026 — Xu & Malkin, “A Penny for Your Prompts: Experiments Detecting and Mitigating LLM Usage by Survey Respondents,” arXiv:2607.00403
• SRC-LD07-YANG-DATA-ANNOTATION-EVALUATION — Yang, Wang, Zhou & Xu, “Data Annotation with Large Language Models: Lessons from a Large Empirical Evaluation,” conditionally accepted at AJPS; pre-publication manuscript
• SRC-LD07-ZHANG-HOMOGENIZATION-2025 — Zhang, Xu & Alvero, “Generative AI Meets Open-Ended Survey Responses,” Sociological Methods & Research 54(3):1197–1242 (2025)
• SRC-LD10-AGARWAL-SEN-2026 — Saharsh Agarwal & Ananya Sen, “The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment,” SSRN 6513059.
• SRC-LD10-CETAS-2024-09 — CETaS / Alan Turing Institute, Sep 2024
• SRC-LD10-CHEN-JEPS-2026 — Chen et al., Journal of Experimental Political Science 2026
• SRC-LD10-COSTELLO-2024 — Costello et al., Science, 2024 — conspiracy-belief experiment
• SRC-LD10-DUFOUR-AMMEBA-ARXIV-2405-11697 — Dufour et al., AMMeBa, arXiv:2405.11697
• SRC-LD10-EBU-BBC-NEWS-INTEGRITY-2025 — EBU/BBC, News Integrity in AI Assistants, Oct 2025
• SRC-LD10-EC-TRANSPARENCY-CODE-2026-06-10 — European Commission, final Transparency Code, 10 Jun 2026
• SRC-LD10-EU-AI-ACT-ART-50 — AI Act Art. 50
• SRC-LD10-GALLEGOS-PNAS-NEXUS-2026 — Gallegos et al., PNAS Nexus 2026
• SRC-LD10-GHOLAMI-SSRN-6238578 — Gholami, Firullo, Cheyre & Acquisti 2026, SSRN 6238578
• SRC-LD10-HACKENBURG-2026 — Hackenburg et al. 2026, arXiv:2604.09200
• SRC-LD10-HACKENBURG-ARXIV-2606-16475 — Hackenburg et al. 2026, arXiv:2606.16475
• SRC-LD10-HACKENBURG-SCIENCE-2025 — Hackenburg et al., Science 2025 / arXiv full text
• SRC-LD10-HOLBLING-MAIER-FEUERRIEGEL-2025 — Hölbling, Maier & Feuerriegel, Scientific Reports 2025
• SRC-LD10-LI-ARAL-2025 — Li & Aral 2025, arXiv:2504.06435
• SRC-LD10-LIN-NATURE-2025-AND-SI — Lin et al., Nature 2025 + Supplementary Information
• SRC-LD10-PEW-2025-07-22 — Pew Research Center, 22 Jul 2025
• SRC-LD10-PNAS-NEXUS-2025-MISLEADING-IMAGE-LABELS — PNAS Nexus 2025 misleading-image labelling study
• SRC-LD10-REUTERS-DNR-2026 — Reuters Institute Digital News Report 2026
• SRC-LD10-REUTERS-DNR-2026-CHAPTER-EXEC-SUMMARY — Reuters Institute DNR 2026 chapter + executive summary
• SRC-LD10-SALVI-2025 — Salvi et al., Nature Human Behaviour, 2025 — personalized GPT-4 persuasion experiment
• SRC-LD10-SALVI-CORRECTION-2026 — Salvi et al. Author Correction, 3 September 2026
• SRC-LD10-SCIENCE-EOC-2026 — Science Editorial Expression of Concern, 11 June 2026
• SRC-LD10-WANG-ARXIV-2608-18352 — Wang et al. 2026, arXiv:2608.18352
• SRC-LD10-XU-IQBAL-MONTGOMERY-2026 — Xu, Iqbal & Montgomery 2026, arXiv:2605.14021
• SRC-LD11-ANTHROPIC-COMPUTE-ANNOUNCEMENTS-2026 — Anthropic 2026 compute announcements
• SRC-LD11-ASML-ANNUAL-REPORT-2025 — ASML 2025 Annual Report
• SRC-LD11-BROADCOM-SEC-8K — Broadcom SEC 8-K
• SRC-LD11-COUNTERPOINT-DRAM-HBM-SHARE-2026-09 — Counterpoint Research, Global DRAM and HBM Market Share, Sep 2026
• SRC-LD11-EC-AI-GIGAFACTORIES-CALL-2026-07 — AI Gigafactories call, 30 Jul 2026
• SRC-LD11-EC-CADA-PROPOSAL-2026-06 — European Commission CADA proposal, 3 Jun 2026
• SRC-LD11-EC-DMA-CLOUD-PRELIMINARY-2026-06 — European Commission, 25 Jun 2026 preliminary DMA position
• SRC-LD11-EPOCH-AI-CHIP-OWNERS-2026-04-14 — Epoch AI, Emberson, You & Somala, 14 Apr 2026
• SRC-LD11-EPOCH-GOOGLE-COMPUTE-2026-04-07 — Epoch AI, 7 Apr 2026
• SRC-LD11-EPOCH-PLUNGING-PRICE-OF-THOUGHT-2026-09 — Epoch AI, The Plunging Price of Thought, Sep 2026
• SRC-LD11-FTC-6B-CSP-AI-PARTNERSHIPS-2025 — FTC Office of Technology 6(b) staff report, Jan 2025
• SRC-LD11-MENLO-STATE-OF-GENAI-ENTERPRISE-2025 — Menlo Ventures, 2025 State of Generative AI in the Enterprise
• SRC-LD11-MICROSOFT-OPENAI-ANNOUNCEMENTS-2025-2026 — Microsoft/OpenAI joint announcements, 28 Oct 2025 and 27 Apr 2026
• SRC-LD11-MUBADALA-LETTER-TO-PARTNERS-2026 — Mubadala 2026 Letter to Partners
• SRC-LD11-NSF-NAIRR-REPORTING-2026 — NSF NAIRR pages and 1 Sep 2026 Operations Center announcement
• SRC-LD11-NVIDIA-10Q-2026-07-26 — NVIDIA Form 10-Q, quarter ended 26 Jul 2026
• SRC-LD11-NVIDIA-Q2-FY2027-REPORTING — NVIDIA Q2 FY2027 reporting and SEC filing
• SRC-LD11-OPENROUTER-A16Z-100T-TOKEN-STUDY — OpenRouter/a16z, State of AI: 100 Trillion Token Study
• SRC-LD11-SYNERGY-CLOUD-2Q26 — Synergy Research Group, 30 Jul 2026
• SRC-LD11-TRENDFORCE-FOUNDRY-2Q26 — TrendForce, 2Q26 foundry release, 9 Sep 2026
• SRC-LD11-UK-CMA-CLOUD-ACTIONS-2026-03 — CMA package of actions, 31 March 2026
• SRC-LD11-UK-CMA-CLOUD-FINAL-DECISION-2025 — UK CMA Cloud Services Market Investigation, final decision 31 Jul 2025
• SRC-LD11-US-COMMERCE-G42-HUMAIN-2025-11 — U.S. Department of Commerce, 19 Nov 2025
• SRC-LD12-AGENCE-EUROPE-RFI-RELAY — Agence Europe relay of Commission statement
• SRC-LD12-CAC-GENAI-FILING-NOTICES — CAC generative-AI filing notices
• SRC-LD12-CEN-CENELEC-EN-18286-2026 — CEN-CENELEC EN 18286 publication
• SRC-LD12-CJEU-C203-22-2025 — CJEU C-203/22, ECLI:EU:C:2025:117
• SRC-LD12-COLORADO-AI-BILLS-2024-2026 — Colorado General Assembly SB24-205, SB25B-004, SB26-189
• SRC-LD12-COLORADO-FISCAL-COURT-ORDER-RECORD — official fiscal/court-order record
• SRC-LD12-COURT-OF-ROME-RG-4785-2025 — Court of Rome R.G. 4785/2025 judgment
• SRC-LD12-DOJ-AI-LITIGATION-TASK-FORCE-MEMO — DOJ AI Litigation Task Force memorandum
• SRC-LD12-EC-AI-ACT-MSA-LIST — European Commission, Market Surveillance Authorities under the AI Act
• SRC-LD12-EC-AI-ACT-STANDARDISATION-GUIDANCE — European Commission AI Act standardisation guidance
• SRC-LD12-EC-AI-GIGAFACTORIES-PAGE — European Commission AI Gigafactories page
• SRC-LD12-EC-AI-OFFICE-ENFORCEMENT-FRAMEWORK — European Commission enforcement framework
• SRC-LD12-EC-GPAI-CODE-AND-GUIDELINES — European Commission GPAI Code and provider guidelines
• SRC-LD12-EU-AI-ACT-2026-1744 — Regulation (EU) 2026/1744; European Commission AI Omnibus implementation material
• SRC-LD12-EUROHPC-AIGF-CALL-2026-07 — EuroHPC JU AIGF call
• SRC-LD12-GARANTE-DECISION-755-2024 — Garante Decision No. 755 of 2 Nov 2024
• SRC-LD12-IAPP-CREDO-2025 — IAPP/Credo AI, AI Governance Profession Report 2025
• SRC-LD12-KOREA-NLIC-AI-BASIC-ACT-DECREE — Korean National Law Information Center
• SRC-LD12-MSIT-SOVEREIGN-AI — South Korean Ministry of Science and ICT official Sovereign AI Foundation Model programme releases
• SRC-LD12-NYS-COMPTROLLER-AUDIT-2024-N-6 — NYS Comptroller Audit 2024-N-6
• SRC-LD12-STATECOUNCIL-AIPLUS-2025-11 — 国务院关于深入实施“人工智能+”行动的意见
• SRC-LD12-US-EO-14365 — Executive Order 14365
• SRC-LD12-US-ITA-KOREA-AI-BASIC-ACT-REPORT — U.S. ITA implementation report
• SRC-LD12-WRIGHT-NULL-COMPLIANCE-2024 — Wright et al., “Null Compliance,” FAccT 2024
• SRC-R03-WANG-MORGENSTERN-DICKERSON-NMI-2025 — Wang et al. NMI 2025
• SRC-R04-AMAS-LEGAL-RECORD-2025 — BVwG W256 2235360-1, decision 1 Sep 2025; prior VwGH/BVwG procedural history
• SRC-R04-HOME-OFFICE-ROUTING-SOURCE-SET — ICIBI 2023 inspection; Home Office transparency/guidance; JCWI/Foxglove litigation record for claimant position
• SRC-R04-HQ-CONTESTABILITY-PROVENANCE-SYNTHESIS — DWP; Home Office; Ada Lovelace; Robodebt synthesis
• SRC-R04-HQ-CROSS-CASE-OVERSIGHT-SYNTHESIS — Cross-case synthesis of Poland; Allegheny; DWP; Austria; Ada Lovelace social-care evidence
• SRC-R04-HQ-SELECTION-BURDEN-SYNTHESIS — DWP; DUO; Toeslagen; Poland legal distinction
• SRC-R05-HQ-CROSS-BATCH-SYNTHESIS — R05 — HQ Cross-Batch Synthesis & Adjudication v0.1
• SRC-R06-HQ-CROSS-BATCH-SYNTHESIS — R06 — HQ Cross-Batch Synthesis & Adjudication v0.1
• SRC-R06-HQ-VERIFICATION-C20 — R06 HQ verification
• SRC-R07-HQ-CROSS-BATCH-SYNTHESIS — R07 — HQ Cross-Batch Synthesis & Adjudication v0.1
• SRC-R08-HQ-CROSS-SOURCE-SYNTHESIS — R08 HQ Batches 01–03 + canonical claim set
• SRC-SL-ABDALLA-ABDALLA-GREY-HOODIE-AIES-2021 — Abdalla & Abdalla, “The Grey Hoodie Project: Big Tobacco, Big Tech, and the Threat on Academic Integrity,” AIES 2021
• SRC-SL-ABDALLA-ET-AL-ELEPHANT-ACL-2023 — Abdalla, Wahle, Ruas, Névéol, Ducel, Mohammad & Fort, “The Elephant in the Room: Analyzing the Presence of Big Tech in Natural Language Processing Research,” ACL 2023
• SRC-SL-ACEMOGLU-ECONOMIC-POLICY-2025 — Acemoglu, “The Simple Macroeconomics of AI,” Economic Policy 40(121), 2025
• SRC-SL-AGARWAL-NAAMAN-VASHISTHA-CHI-2025 — Agarwal, Naaman & Vashistha, “AI Suggestions Homogenize Writing Toward Western Styles and Diminish Cultural Nuances,” CHI 2025, arXiv:2409.11360
• SRC-SL-AHMED-DAS-MARTIN-BANERJEE-ARXIV-2405-12193 — Ahmed, Das, Martin & Banerjee, “The Narrow Depth and Breadth of Corporate Responsible AI Research,” arXiv:2405.12193, 2026
• SRC-SL-AHMED-WAHED-ARXIV-2010-15581 — Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020
• SRC-SL-AI-INDEX-2026-ANNUAL-REPORT — Sajadieh, Fattorini, Perrault, Gil, Parli et al., “The AI Index 2026 Annual Report,” AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, April 2026
• SRC-SL-AITKEN-ET-AL-ACL-2024 — Aitken, Abdalla, Rudie & Stinson, “Collaboration or Corporate Capture? Quantifying NLP’s Reliance on Industry Artifacts and Contributions,” ACL 2024
• SRC-SL-ANDERSON-SHAH-KREMINSKI-CC-2024 — Anderson, Shah & Kreminski, “Homogenization Effects of Large Language Models on Human Creative Ideation,” ACM Creativity & Cognition 2024, arXiv:2402.01536
• SRC-SL-ANTHROPIC-BUILDING-EFFECTIVE-AGENTS-2024-12-19 — Anthropic, “Building effective agents,” Anthropic engineering post, 19 December 2024, as served on 16 October 2026
• SRC-SL-ANTHROPIC-ECONOMIC-ADVISORY-COUNCIL-2025-04-28 — Anthropic, “Introducing the Anthropic Economic Advisory Council,” 28 April 2025 (updated 9 May 2025), as served on 16 October 2026
• SRC-SL-ANTHROPIC-ECONOMIC-FUTURES-FUND-2026-07-22 — Anthropic, “A research agenda for the Economic Futures Research Fund,” 22 July 2026, as served on 16 October 2026
• SRC-SL-ANTHROPIC-ECONOMIC-FUTURES-PROGRAM-2026 — Anthropic, “Anthropic Economic Futures Program,” programme page, as served on 16 October 2026
• SRC-SL-ANTHROPIC-ECONOMIC-INDEX-2025-09-15 — Anthropic, “Anthropic Economic Index report: Uneven geographic and enterprise AI adoption,” 15 September 2025
• SRC-SL-ANTHROPIC-ECONOMIC-INDEX-2026-01-15 — Anthropic, “Anthropic Economic Index report: Economic primitives,” 15 January 2026
• SRC-SL-ANTHROPIC-EXTERNAL-RESEARCHER-ACCESS-2026 — Anthropic, “External Researcher Access Program,” Claude help centre page, as served on 16 October 2026
• SRC-SL-ASHKINAZE-ET-AL-CI-2025 — Ashkinaze, Mendelsohn, Li Qiwei, Budak & Gilbert, “How AI Ideas Affect the Creativity, Diversity, and Evolution of Human Ideas: Evidence From a Large, Dynamic Experiment,” ACM Collective Intelligence 2025, arXiv:2401.13481
• SRC-SL-AUTOR-ET-AL-NBER-35720 — Autor, Rodchenko, Martin, Iscenko, Strand, Pearl & Ferere, “Does AI Assistance Enhance or Erode Expertise? Evidence from a Three-Month Field Experiment in Patent Drafting,” NBER Working Paper 35720, 2026
• SRC-SL-BAK-COLEMAN-ET-AL-ARXIV-2510-19894 — Bak-Coleman, O’Connor, Bergstrom & West, “The Risks of Industry Influence in Tech Research,” arXiv:2510.19894, 2025
• SRC-SL-BASLANDZE-ET-AL-NBER-34984 — Baslandze, Edwards, Graham et al., “Artificial Intelligence, Productivity, and the Workforce: Evidence from Corporate Executives,” NBER Working Paper 34984, March 2026
• SRC-SL-BASTANI-ET-AL-PNAS-2025 — Bastani, Bastani, Sungu, Ge, Kabakcı & Mariman, “Generative AI without guardrails can harm learning: Evidence from high school mathematics,” PNAS 122(26), 2025
• SRC-SL-BAVARESCO-ET-AL-JUDGE-BENCH-ACL-2025 — Bavaresco, Bernardi, Bertolazzi, Elliott, Fernández, Gatt et al., “LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Tasks,” ACL 2025, arXiv:2406.18403
• SRC-SL-BEAN-ET-AL-NEURIPS-2025-ARXIV-2511-04703 — Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703
• SRC-SL-BEANE-ASQ-2019 — Beane, “Shadow Learning: Building Robotic Surgical Skill When Approved Means Fail,” Administrative Science Quarterly 64(1), 2019
• SRC-SL-BECKER-METR-SELF-REPORT-SURVEY-2026-05-11 — Becker, “Measuring the Self-Reported Impact of Early-2026 AI on Technical Worker Productivity,” METR, 11 May 2026
• SRC-SL-BECKER-RUSH-BARNES-REIN-ARXIV-2507-09089 — Becker, Rush, Barnes & Rein, “Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity,” arXiv:2507.09089, July 2025
• SRC-SL-BERGLUND-ET-AL-REVERSAL-ARXIV-2309-12288 — Berglund, Tong, Kaufmann, Balesni, Cooper Stickland, Korbak & Evans, “The Reversal Curse: LLMs trained on ‘A is B’ fail to learn ‘B is A’,” ICLR 2024
• SRC-SL-BESIROGLU-ET-AL-ARXIV-2401-02452 — Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024
• SRC-SL-BICK-BLANDIN-DEMING-NBER-W32966 — Bick, Blandin & Deming, “The Rapid Adoption of Generative AI,” NBER Working Paper 32966, 2024 (revised February 2025)
• SRC-SL-BICK-BLANDIN-DEMING-SCHUMACHER-STLFED-WP-2026-017 — Bick, Blandin, Deming & Schumacher, “What Work Does Generative AI Do?,” Federal Reserve Bank of St. Louis Working Paper 2026-017A, August 2026
• SRC-SL-BICK-BLANDIN-DEMING-STLFED-WP-2024-027F — Bick, Blandin & Deming, “The Rapid Adoption of Generative AI,” Federal Reserve Bank of St. Louis Working Paper 2024-027F, October 2025
• SRC-SL-BICK-ET-AL-STLFED-WP-2026-003B — Bick, Blandin, Deming, Fuchs-Schündeln & Jessen, “Mind the Gap: AI Adoption in Europe and the US,” Federal Reserve Bank of St. Louis Working Paper 2026-003B, 2026
• SRC-SL-BIRHANE-ET-AL-AI-AUDITING-SATML-2024 — Birhane, Steed, Ojewale, Vecchione & Raji, “AI auditing: The Broken Bus on the Road to AI Accountability,” IEEE SaTML 2024, arXiv:2401.14462
• SRC-SL-BIRHANE-ET-AL-VALUES-FACCT-2022 — Birhane, Kalluri, Card, Agnew, Dotan & Bao, “The Values Encoded in Machine Learning Research,” FAccT 2022
• SRC-SL-BIS-AER-2024-CHAPTER-III — Bank for International Settlements, “Artificial intelligence and the economy: implications for central banks,” Annual Economic Report 2024, chapter III
• SRC-SL-BIS-QUARTERLY-REVIEW-2025-12 — Bank for International Settlements, BIS Quarterly Review, December 2025, overview (“Volatility challenges risk-taking”)
• SRC-SL-BOMMASANI-ET-AL-ARXIV-2108-07258 — Bommasani et al., “On the Opportunities and Risks of Foundation Models,” arXiv:2108.07258, 2021
• SRC-SL-BOMMASANI-ET-AL-HIRING-FACCT-2026 — Bommasani, Bana, Creel, Jurafsky & Liang, “Algorithmic Monocultures in Hiring,” FAccT 2026, arXiv:2605.27371
• SRC-SL-BRYNJOLFSSON-HITT-JEP-2000 — Brynjolfsson & Hitt, “Beyond Computation: Information Technology, Organizational Transformation and Business Performance,” Journal of Economic Perspectives 14(4), 2000
• SRC-SL-BRYNJOLFSSON-LI-RAYMOND-ARXIV-2304-11771-V2 — Brynjolfsson, Li & Raymond, “Generative AI at Work,” arXiv:2304.11771 v2, November 2024
• SRC-SL-BRYNJOLFSSON-ROCK-SYVERSON-AEJ-MACRO-2021 — Brynjolfsson, Rock & Syverson, “The Productivity J-Curve: How Intangibles Complement General Purpose Technologies,” American Economic Journal: Macroeconomics 13(1), 2021
• SRC-SL-BUDZYN-ET-AL-LANCET-GH-2025 — Budzyń et al., “Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study,” Lancet Gastroenterology & Hepatology 10(10), 2025
• SRC-SL-BURTCH-LEE-CHEN-SCI-REP-2024 — Burtch, Lee & Chen, “The consequences of generative AI for online knowledge communities,” Scientific Reports 14, 10413, 2024
• SRC-SL-BURTCH-MONOCULTURE-ARXIV-2607-13077 — Burtch, “The Hitchhiker’s Guide to Monoculture: AI Homogenizes Syntax, Not (Necessarily) Semantics,” arXiv:2607.13077, 2026
• SRC-SL-CHAN-ET-AL-AGENTIC-ARXIV-2302-10329 — Chan et al., “Harms from Increasingly Agentic Algorithmic Systems,” FAccT 2023
• SRC-SL-CHATTERJI-ET-AL-NBER-W34255 — Chatterji, Cunningham, Deming, Hitzig, Ong, Shan & Wadman, “How People Use ChatGPT,” NBER Working Paper 34255, 2025
• SRC-SL-CHELIKAVADA-BENNETT-ARXIV-2509-11982 — Chelikavada & Bennett, “Examining the Relationship between Scientific Publishing Activity and Hype-Driven Financial Bubbles: A Comparison of the Dot-Com and AI Eras,” arXiv:2509.11982v2, 2025
• SRC-SL-CHEN-GASCO-HERNANDEZ-ESTEVE-ARPA-2024 — Chen, Gascó-Hernandez & Esteve, “The Adoption and Implementation of Artificial Intelligence Chatbots in Public Organizations: Evidence from U.S. State Governments,” American Review of Public Administration 54(3), 2024
• SRC-SL-COLOMBATTO-FLEMING-NEUROSCI-CONSCIOUSNESS-2024 — Colombatto & Fleming, “Folk psychological attributions of consciousness to large language models,” Neuroscience of Consciousness, 2024
• SRC-SL-COSTANZA-CHOCK-ET-AL-FACCT-2022 — Costanza-Chock, Harvey, Raji, Czernuszenko & Buolamwini, “Who Audits the Auditors? Recommendations from a field scan of the algorithmic auditing ecosystem,” FAccT 2022, arXiv:2310.02521
• SRC-SL-CRUCES-ET-AL-ARXIV-2608-04198 — Cruces, Fernández Meijide, Galiani, Gálvez & Lombardi, “Does generative AI narrow education-based productivity gaps? Evidence from a randomized experiment,” arXiv:2608.04198, submitted 4 August 2026
• SRC-SL-CUNNINGHAM-WHITFILL-METR-2026-05-08 — Cunningham & Whitfill, “Task Substitution and Uplift,” METR, 8 May 2026
• SRC-SL-DAVIS-DALLAS-FED-2026-02-24 — Davis, “AI is simultaneously aiding and replacing workers, wage data suggest,” Federal Reserve Bank of Dallas, 24 February 2026
• SRC-SL-DEL-RIO-CHANONA-ET-AL-PNAS-NEXUS-2024 — del Rio-Chanona, Laurentsyeva & Wachs, “Large language models reduce public knowledge sharing on online Q&A platforms,” PNAS Nexus 3, pgae400, 2024
• SRC-SL-DELLACQUA-ET-AL-ORGANIZATION-SCIENCE-37-4-2026 — Dell’Acqua, Ayoubi, Lifshitz et al., “The Cybernetic Teammate: A Field Experiment on Generative AI and Teamwork,” Organization Science 37(4), 2026
• SRC-SL-DEMIRER-FRADKIN-TADELIS-PENG-NBER-34608 — Demirer, Fradkin, Tadelis & Peng, “The Emerging Market for Intelligence: Pricing, Supply, and Demand for LLMs,” NBER Working Paper 34608, December 2025
• SRC-SL-DILLON-JAFFE-IMMORLICA-STANTON-ARXIV-2504-11436-V4 — Dillon, Jaffe, Immorlica & Stanton, “Shifting Work Patterns with Generative AI,” NBER Working Paper 33795, revised November 2025
• SRC-SL-DOSHI-HAUSER-SCIENCE-ADVANCES-2024 — Doshi & Hauser, “Generative AI enhances individual creativity but reduces the collective diversity of novel content,” Science Advances 10, eadn5290, 2024
• SRC-SL-DZIRI-ET-AL-FAITH-FATE-ARXIV-2305-18654 — Dziri et al., “Faith and Fate: Limits of Transformers on Compositionality,” NeurIPS 2023
• SRC-SL-EC-DSA-DELEGATED-ACT-DATA-ACCESS-2025-07-02 — European Commission, “Commission adopts delegated act on data access under the Digital Services Act,” 2 July 2025, as served on 16 October 2026
• SRC-SL-EC-DSA-VETTED-RESEARCHERS-ROUNDTABLE-2026-05-20 — European Commission, “Commission holds roundtable on data access for vetted researchers,” 20 May 2026, as served on 16 October 2026
• SRC-SL-EC-GPAI-GUIDELINES-C-2025-7719 — European Commission, “Guidelines on the scope of the obligations for providers of general-purpose AI models established by Regulation (EU) 2024/1689 (AI Act),” C(2025) 7719 final, 19 November 2025
• SRC-SL-EC-X-DSA-ACTION-PLAN-2026-07-16 — European Commission, “Commission accepts X’s action plan to comply with Digital Services Act,” 16 July 2026, as served on 16 October 2026
• SRC-SL-EC-X-DSA-FINE-2025-12-05 — European Commission, “Commission fines X €120 million under the Digital Services Act,” press release IP/25/2934, 5 December 2025
• SRC-SL-ELOUNDOU-ET-AL-ARXIV-2303-10130 — Eloundou, Manning, Mishkin & Rock, “GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models,” arXiv:2303.10130, 2023
• SRC-SL-ESPOSITO-MIT-PRESS-2022 — Esposito, Artificial Communication: How Algorithms Produce Social Intelligence, MIT Press, 2022
• SRC-SL-EU-GPAI-CODE-OF-PRACTICE-SAFETY-SECURITY-2025-07-10 — European Commission, “The General-Purpose AI Code of Practice: Safety and Security chapter,” final version, 10 July 2025
• SRC-SL-EU-REG-2026-1744 — Regulation (EU) 2026/1744 (Digital Omnibus on AI), 8 Jul 2026
• SRC-SL-EVANS-SCIENCE-2008 — Evans, “Electronic Publication and the Narrowing of Science and Scholarship,” Science 321, 2008
• SRC-SL-FANG-YUAN-ZHANG-DONATI-SARVARY-ARXIV-2510-12049-V6 — Fang, Yuan, Zhang, Donati & Sarvary, “Generative AI and Sales Productivity: Field Experiments in Online Retail,” arXiv:2510.12049, v6, June 2026
• SRC-SL-FEFFER-ET-AL-RED-TEAMING-AIES-2024 — Feffer, Sinha, Deng, Lipton & Heidari, “Red-Teaming for Generative AI: Silver Bullet or Security Theater?,” AIES 2024, arXiv:2401.15897
• SRC-SL-FENOGLIO-ARXIV-2607-28137 — Fenoglio, “Asymmetric Communication: Large Language Models and Language Games,” arXiv:2607.28137, 2026
• SRC-SL-GAMBACORTA-KHARROUBI-MEHROTRA-OLIVIERO-BIS-1321 — Gambacorta, Kharroubi, Mehrotra & Oliviero, “Artificial intelligence and growth in advanced and emerging economies: short-run impact,” BIS Working Papers No 1321, December 2025
• SRC-SL-GAO-WANG-NHB-2024 — Gao & Wang, “Quantifying the use and potential benefits of artificial intelligence in scientific research,” Nature Human Behaviour 8, 2024 (arXiv:2304.10578)
• SRC-SL-GB-T-45654-2025-CSET-TRANSLATION — State Administration for Market Regulation & Standardization Administration of China, GB/T 45654-2025, “Cybersecurity Technology — Basic Safety Requirements for Generative Artificial Intelligence Services,” issued 25 April 2025 (CSET English translation, 28 May 2026)
• SRC-SL-GEMA-ET-AL-ARXIV-2406-04127 — Gema, Leang, Hong et al., “Are We Done with MMLU?,” arXiv:2406.04127 (v3), January 2025
• SRC-SL-GERSTGRASSER-ET-AL-ARXIV-2404-01413 — Gerstgrasser, Schaeffer, Dey, Rafailov, Sleight, Hughes et al., “Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data,” arXiv:2404.01413, 2024
• SRC-SL-GNEWUCH-ET-AL-ARXIV-2512-05714 — Gnewuch, Wahle, Ruas & Gipp, “Big Tech-Funded AI Papers Have Higher Citation Impact, Greater Insularity, and Larger Recency Bias,” arXiv:2512.05714, 2025
• SRC-SL-GREEN-JSC-2021 — Green, “The Contestation of Tech Ethics: A Sociotechnical Approach to Technology Ethics in Practice,” Journal of Social Computing, 2021
• SRC-SL-GURNEE-TEGMARK-ARXIV-2310-02207 — Gurnee & Tegmark, “Language Models Represent Space and Time,” ICLR 2024
• SRC-SL-HAGENDORFF-MEDING-AI-SOCIETY-2021 — Hagendorff & Meding, “Ethical considerations and statistical analysis of industry involvement in machine learning research,” AI & Society, 2021
• SRC-SL-HANDA-ET-AL-ARXIV-2503-04761 — Handa, Tamkin, McCain, Huang, Durmus et al., “Which Economic Tasks are Performed with AI? Evidence from Millions of Claude Conversations,” arXiv:2503.04761, 2025
• SRC-SL-HAO-XU-LI-EVANS-NATURE-2026 — Hao, Xu, Li & Evans, “Artificial intelligence tools expand scientists’ impact but contract science’s focus,” Nature, 2026
• SRC-SL-HARDY-ET-AL-MORE-THAN-MARKETING-ARXIV-2412-05520 — Hardy, Reuel, Jafari Meimandi, Soder, Griffith, Asmar, Koyejo, Bernstein & Kochenderfer, “More than Marketing? On the Information Value of AI Benchmarks for Practitioners,” arXiv:2412.05520, December 2024
• SRC-SL-HASLBERGER-GINGRICH-BHATIA-SSRN-4594466 — Haslberger, Gingrich & Bhatia, “No Great Equalizer: Experimental Evidence on Productivity Effects of Generative AI Use in the UK Labor Market,” SSRN Working Paper 4594466, 2023
• SRC-SL-HE-MILLER-AGARWAL-KASTNER-VASILESCU-MSR-2026 — He, Miller, Agarwal, Kästner & Vasilescu, “Speed at the Cost of Quality: How Cursor AI Increases Short-Term Velocity and Long-Term Complexity in Open-Source Projects,” Proceedings of MSR 2026
• SRC-SL-HEDDEN-RAGHAVAN-PHIL-PERSPECTIVES-2026 — Hedden & Raghavan, “Algorithmic monoculture and its critics,” Philosophical Perspectives, 2026 (arXiv:2604.06047)
• SRC-SL-HOSSEINI-MAASOUM-LICHTINGER-SSRN-5425555 — Hosseini Maasoum & Lichtinger, “Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data,” SSRN Working Paper 5425555, version of 6 June 2026
• SRC-SL-HUANG-JIN-LI-ARXIV-2412-03963 — Huang, Jin & Li, “Augmenting Minds or Automating Skills: The Differential Role of Human Capital in Generative AI’s Impact on Creative Tasks,” arXiv:2412.03963, December 2024
• SRC-SL-IDAN-ANAND-ARXIV-2605-18143 — Idan & Anand, “Generative AI and the Productivity Divide: Human-AI Complementarities in Education,” arXiv:2605.18143, May 2026
• SRC-SL-IDE-ARXIV-2507-16078-V9 — Ide, “Automation, AI, and the Intergenerational Transmission of Knowledge,” arXiv:2507.16078 v9, June 2026
• SRC-SL-IDE-TALAMAS-JPE — Ide & Talamàs, “Artificial Intelligence in the Knowledge Economy,” Journal of Political Economy, doi:10.1086/737233 (abstract read in arXiv:2312.05481 v12, May 2025)
• SRC-SL-INTERNATIONAL-AI-SAFETY-REPORT-2026 — Bengio, Clare, Prunkl, Andriushchenko, Bucknall, Murray et al., “International AI Safety Report 2026,” arXiv:2602.21012, February 2026
• SRC-SL-JO-GARG-RAGHAVAN-ARXIV-2602-24086 — Jo, Garg & Raghavan, “The Subjectivity of Monoculture,” arXiv:2602.24086, 2026
• SRC-SL-JOBIN-IENCA-VAYENA-NMI-2019 — Jobin, Ienca & Vayena, “The global landscape of AI ethics guidelines,” Nature Machine Intelligence, 2019
• SRC-SL-JUROWETZKI-ET-AL-ARXIV-2102-01648 — Jurowetzki, Hain, Mateos-Garcia & Stathoulopoulos, “The Privatization of AI Research(-ers): Causes and Potential Consequences,” arXiv:2102.01648, 2021
• SRC-SL-KASIRZADEH-GABRIEL-ARXIV-2504-21848 — Kasirzadeh & Gabriel, “Characterizing AI Agents for Alignment and Governance,” arXiv:2504.21848, 2025
• SRC-SL-KIM-ET-AL-CORRELATED-ERRORS-ICML-2025 — Kim, Garg, Peng & Garg, “Correlated Errors in Large Language Models,” ICML 2025, arXiv:2506.07962
• SRC-SL-KLEINBERG-RAGHAVAN-PNAS-2021 — Kleinberg & Raghavan, “Algorithmic monoculture and social welfare,” PNAS 118, 2021 (arXiv:2101.05853)
• SRC-SL-KLINGER-ET-AL-ARXIV-2009-10385 — Klinger, Mateos-Garcia & Stathoulopoulos, “A narrowing of AI research?”, arXiv:2009.10385, 2022
• SRC-SL-KOLT-GOVERNING-AI-AGENTS-ARXIV-2501-07913 — Kolt, “Governing AI Agents,” arXiv:2501.07913, 2025 (forthcoming in the Notre Dame Law Review)
• SRC-SL-KROKOWSKI-HIRSCH-KREINSEN-AI-SOCIETY-2025 — Krokowski & Hirsch-Kreinsen, “Once promised, forever believed? An essay on the blind spots of AI promises,” AI & Society, 2025
• SRC-SL-KUMAR-ET-AL-CHI-2025 — Kumar, Vincentius, Jordan & Anderson, “Human Creativity in the Age of LLMs: Randomized Experiments on Divergent and Convergent Thinking,” CHI 2025, arXiv:2410.03703
• SRC-SL-KUSUMEGI-ET-AL-REPLY-ARXIV-2607-28968 — Kusumegi et al., “A robust association between LLM use and scientific productivity: Assessing stopping-time selection,” arXiv:2607.28968, July 2026
• SRC-SL-KUSUMEGI-ET-AL-SCIENCE-2025 — Kusumegi, Yang, Ginsparg, de Vaan, Stuart & Yin, “Scientific production in the era of large language models,” Science 390, 2025 (author version arXiv:2601.13187)
• SRC-SL-KWA-ET-AL-METR-ARXIV-2503-14499 — Kwa et al., “Measuring AI Ability to Complete Long Software Tasks,” arXiv:2503.14499v4, 2026
• SRC-SL-LAUFER-ET-AL-FACCT-2022 — Laufer, Jain, Cooper, Kleinberg & Heidari, “Four Years of FAccT: A Reflexive, Mixed-Methods Analysis of Research Contributions, Shortcomings, and Future Prospects,” FAccT 2022
• SRC-SL-LEBOVITZ-LIFSHITZ-ASSAF-LEVINA-ORGANIZATION-SCIENCE-2022 — Lebovitz, Lifshitz-Assaf & Levina, “To Engage or Not to Engage with AI for Critical Judgments: How Professionals Deal with Opacity When Using AI for Medical Diagnosis,” Organization Science 33(1), 2022
• SRC-SL-LEHMANN-CORNELIUS-STING-ARXIV-2409-09047 — Lehmann, Cornelius & Sting, “AI Meets the Classroom: When Do Large Language Models Harm Learning?,” arXiv:2409.09047 v2, March 2025
• SRC-SL-LI-ET-AL-OTHELLO-ARXIV-2210-13382 — Li, Hopkins, Bau, Viégas, Pfister & Wattenberg, “Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task,” ICLR 2023
• SRC-SL-LIANG-ET-AL-ARXIV-2404-01268 — Liang, Zhang, Wu, Lepp, Ji, Zhao et al., “Mapping the Increasing Use of LLMs in Scientific Papers,” arXiv:2404.01268, April 2024
• SRC-SL-LMARENA-RESPONSE-LEADERBOARD-ILLUSION-2025-05-09 — LMArena, “Our Response to ‘The Leaderboard Illusion’ Writeup,” Arena blog, 9 May 2025 (updated 13 June 2025)
• SRC-SL-LODEFALK-LOTHMAN-KOCH-ENGBERG-ORU-WP-2-2026 — Lodefalk, Löthman, Koch & Engberg, “Same Storm, Different Boats: Generative AI and the Age Gradient in Hiring,” Örebro University Working Paper 2/2026, version of 16 March 2026
• SRC-SL-LONGPRE-ET-AL-SAFE-HARBOR-ARXIV-2403-04893 — Longpre, Kapoor, Klyman, Ramaswami, Bommasani et al., “A Safe Harbor for AI Evaluation and Red Teaming,” arXiv:2403.04893, 2024
• SRC-SL-LUITSE-DENKENA-BDS-2021 — Luitse & Denkena, “The great Transformer: Examining the role of large language models in the political economy of AI,” Big Data & Society, 2021
• SRC-SL-MA-ET-AL-WILDCHAT-AUSTRALIA-ARXIV-2609-28990 — Ma, Gero, Canonne, Jin & Thilakarathna, “How People Use ChatGPT in Australia: A WildChat Analysis,” arXiv:2609.28990, 2026
• SRC-SL-MAIA-POLO-ET-AL-TINYBENCHMARKS-ICML-2024 — Maia Polo, Weber, Choshen, Sun, Xu & Yurochkin, “tinyBenchmarks: evaluating LLMs with fewer examples,” ICML 2024, arXiv:2402.14992
• SRC-SL-MAIER-ET-AL-ARXIV-2605-04779 — Maier, Gunzenhäuser, Schweisthal, Schneider & Feuerriegel, “A meta-analysis of the effect of generative AI on productivity and learning in programming,” arXiv:2605.04779, May 2026
• SRC-SL-MARKUS-GILARDI-STETSENKO-ARXIV-2506-08738 — Markus, Gilardi & Stetsenko, “Societal AI Research Has Become Less Interdisciplinary,” arXiv:2506.08738, 2025
• SRC-SL-MCCOY-ET-AL-EMBERS-ARXIV-2309-13638 — McCoy, Yao, Friedman, Hardy & Griffiths, “Embers of Autoregression: Understanding Large Language Models Through the Problem They are Trained to Solve,” arXiv:2309.13638, 2023
• SRC-SL-MCCOY-ET-AL-O1-ARXIV-2410-01792 — McCoy, Yao, Friedman, Hardy & Griffiths, “When a language model is optimized for reasoning, does it still show embers of autoregression? An analysis of OpenAI o1,” arXiv:2410.01792, 2024
• SRC-SL-MEINCKE-NAVE-TERWIESCH-CHATGPT-IDEA-DIVERSITY — Meincke, Nave & Terwiesch, “ChatGPT Decreases Idea Diversity in Brainstorming,” author manuscript, The Wharton School, University of Pennsylvania
• SRC-SL-MESSERI-CROCKETT-NATURE-2024 — Messeri & Crockett, “Artificial intelligence and illusions of understanding in scientific research,” Nature 627, 2024
• SRC-SL-METR-DEVELOPER-STUDY-DESIGN-2026-02-24 — METR, “We are Changing our Developer Productivity Experiment Design,” 24 February 2026
• SRC-SL-METR-TIME-HORIZON-1-1-2026-01-29 — METR, “Time Horizon 1.1,” METR blog post, 29 January 2026
• SRC-SL-MICROSOFT-AIEI-PROGRAM-2025 — Microsoft Research, “AI Economy Institute (AIEI) Program,” call for proposals, 2025, as served on 16 October 2026
• SRC-SL-MILLER-ET-AL-ICSE-2026 — Miller, Choudhuri, Ulloa et al., “‘Maybe We Need Some More Examples:’ Individual and Team Drivers of Developer GenAI Tool Use,” Proceedings of ICSE 2026
• SRC-SL-MILLIERE-BUCKNER-ARXIV-2401-03910 — Millière & Buckner, “A Philosophical Introduction to Language Models — Part I: Continuity With Classic Debates,” arXiv:2401.03910, 2024
• SRC-SL-MIRZADEH-ET-AL-GSM-SYMBOLIC-ARXIV-2410-05229 — Mirzadeh, Alizadeh, Shahrokhi, Tuzel, Bengio & Farajtabar, “GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models,” ICLR 2025
• SRC-SL-MITCHELL-KRAKAUER-ARXIV-2210-13966 — Mitchell & Krakauer, “The Debate Over Understanding in AI’s Large Language Models,” arXiv:2210.13966, 2023
• SRC-SL-MITIAEVA-XIAO-ARXIV-2609-21401 — Mitiaeva & Xiao, “Talking past the machine: morality, politeness, and alignment in human–AI dialogue,” arXiv:2609.21401, 2026
• SRC-SL-MURTFELDT-ET-AL-ARXIV-2404-07340 — Murtfeldt, Paik, Alterman, Kahveci & West, “RIP Twitter API: A Eulogy to Its Vast Research Contributions,” arXiv:2404.07340, 2025
• SRC-SL-NARAYANAN-KAPOOR-KNIGHT-2025-04-15 — Narayanan & Kapoor, “AI as Normal Technology: An alternative to the vision of AI as a potential superintelligence,” Knight First Amendment Institute, 15 April 2025
• SRC-SL-NEEDHAM-ET-AL-EVALUATION-AWARENESS-ARXIV-2505-23836 — Needham, Edkins, Pimpale, Bartsch & Hobbhahn, “Large Language Models Often Know When They Are Being Evaluated,” arXiv:2505.23836 (v3), July 2025
• SRC-SL-NEUMANN-ET-AL-XP-2026 — Neumann, Bischof, Hinz et al., “Between Policy and Practice: GenAI Adoption in Agile Software Development Teams,” XP 2026, LNBIP
• SRC-SL-NOY-ZHANG-SCIENCE-2023 — Noy & Zhang, “Experimental evidence on the productivity effects of generative artificial intelligence,” Science 381(6654), 2023
• SRC-SL-OTIS-CLARKE-DELECOURT-HOLTZ-KONING-MNSC-2026 — Otis, Clarke, Delecourt, Holtz & Koning, “The Uneven Impact of Generative AI on Entrepreneurial Performance: Evidence from a Field Experiment in Kenya,” Management Science, 2026
• SRC-SL-PADMAKUMAR-HE-ICLR-2024 — Padmakumar & He, “Does Writing with Language Models Reduce Content Diversity?,” ICLR 2024, arXiv:2309.05196
• SRC-SL-REIN-METR-ALGORITHMIC-HOLISTIC-2025-08-13 — Rein, “Research Update: Algorithmic vs. Holistic Evaluation,” METR, 13 August 2025
• SRC-SL-RENAULT-BERGEAUD-BOSQUET-ARXIV-2605-17979 — Renault, Bergeaud & Bosquet, “Comment on Scientific production in the era of large language models,” arXiv:2605.17979, May 2026
• SRC-SL-REUEL-ET-AL-BETTERBENCH-ARXIV-2411-12990 — Reuel, Hardy, Smith, Lamparth, Hardy & Kochenderfer, “BetterBench: Assessing AI Benchmarks, Uncovering Issues, and Establishing Best Practices,” NeurIPS 2024, arXiv:2411.12990
• SRC-SL-ROY-CHOWDHURY-GARIMELLA-ARXIV-2610-18465 — Roy Chowdhury & Garimella, “How Safe Is ChatGPT Health Advice? Evidence from Real User Conversations in Four Global South Countries,” arXiv:2610.18465, 2026
• SRC-SL-SCHOEN-ET-AL-ANTI-SCHEMING-ARXIV-2509-15541 — Schoen, Nitishinskaya, Balesni, Højmark, Hofstätter, Scheurer et al., “Stress Testing Deliberative Alignment for Anti-Scheming Training,” arXiv:2509.15541, September 2025
• SRC-SL-SCLAR-CHOI-TSVETKOV-SUHR-ICLR-2024 — Sclar, Choi, Tsvetkov & Suhr, “Quantifying Language Models’ Sensitivity to Spurious Features in Prompt Design,” ICLR 2024, arXiv:2310.11324
• SRC-SL-SHANAHAN-MCDONELL-REYNOLDS-ARXIV-2305-16367 — Shanahan, McDonell & Reynolds, “Role-Play with Large Language Models,” arXiv:2305.16367, 2023
• SRC-SL-SHAO-ET-AL-WORKBANK-ARXIV-2506-06576 — Shao, Zope, Jiang, Pei, Nguyen, Brynjolfsson & Yang, “Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce,” arXiv:2506.06576, 2026
• SRC-SL-SHEN-TAMKIN-ARXIV-2601-20245 — Shen & Tamkin, “How AI Impacts Skill Formation,” arXiv:2601.20245 v2, February 2026
• SRC-SL-SHUMAILOV-ET-AL-NATURE-2024 — Shumailov, Shumaylov, Zhao, Papernot, Anderson & Gal, “AI models collapse when trained on recursively generated data,” Nature 631, 2024
• SRC-SL-SI-HASHIMOTO-YANG-ARXIV-2506-20803 — Si, Hashimoto & Yang, “The Ideation–Execution Gap: Execution Outcomes of LLM-Generated versus Human Research Ideas,” arXiv:2506.20803, June 2025
• SRC-SL-SI-YANG-HASHIMOTO-ICLR-2025 — Si, Yang & Hashimoto, “Can LLMs Generate Novel Research Ideas? A Large-Scale Human Study with 100+ NLP Researchers,” ICLR 2025, arXiv:2409.04109
• SRC-SL-SINGH-ET-AL-GLOBAL-MMLU-ARXIV-2412-03304 — Singh, Romanou, Fourrier, Adelani, Ngui, Vila-Suero et al., “Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation,” arXiv:2412.03304 (v2), 2025
• SRC-SL-SINGH-ET-AL-LEADERBOARD-ILLUSION-ARXIV-2504-20879 — Singh, Nan, Wang, D’Souza, Kapoor, Üstün, Koyejo, Deng, Longpre, Smith, Ermis, Fadaee & Hooker, “The Leaderboard Illusion,” arXiv:2504.20879 (v2), May 2025
• SRC-SL-STATE-COUNCIL-AI-PLUS-OPINIONS-2025-08-26 — State Council of the People’s Republic of China, “Opinions on Deepening the Implementation of the ‘Artificial Intelligence+’ Initiative,” 26 August 2025 (English translation by CSET)
• SRC-SL-TAMKIN-ET-AL-CLIO-ARXIV-2412-13678 — Tamkin, McCain, Handa, Durmus et al., “Clio: Privacy-Preserving Insights into Real-World AI Use,” arXiv:2412.13678, 2024
• SRC-SL-TANGI-RODRIGUEZ-MULLER-COMBETTO-EPART-2025 — Tangi, Rodriguez Müller & Combetto, “A Silent Partner: The Shadow Presence of Generative Artificial Intelligence in Public Administrations,” Electronic Participation (ePart 2025), LNCS
• SRC-SL-TAYLOR-ET-AL-SANDBAGGING-AUDITING-GAMES-ARXIV-2512-07810 — Taylor, Black, Bowen, Read, Golechha et al., “Auditing Games for Sandbagging,” arXiv:2512.07810, December 2025
• SRC-SL-TERWIESCH-ET-AL-ARXIV-2607-27553 — Terwiesch, Meincke, Girotra, Mollick, Nave & Ulrich, “AI and Its Impact on Creativity and Diversity: An Empirical Study of LLM-Generated Product Ideas,” arXiv:2607.27553, 2026
• SRC-SL-TOMLINSON-ET-AL-ARXIV-2507-07935 — Tomlinson, Jaffe, Wang, Counts & Suri, “Working with AI: Measuring the Applicability of Generative AI to Occupations,” arXiv:2507.07935, 2025
• SRC-SL-TOUPS-ET-AL-NEURIPS-2023 — Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862
• SRC-SL-TRUONG-ET-AL-ARXIV-2511-16842 — Truong, Tu, Hardy, Reuel, Tang, Burapacheep, Perera, Uwakwe, Domingue, Haber & Koyejo, “Fantastic Bugs and Where to Find Them in AI Benchmarks,” arXiv:2511.16842, November 2025
• SRC-SL-US-UK-AISI-O1-PREDEPLOYMENT-EVALUATION-2024-12 — US AI Safety Institute & UK AI Safety Institute, “US AISI and UK AISI Joint Pre-Deployment Test: OpenAI o1,” December 2024
• SRC-SL-VAN-DER-VLIST-HELMOND-FERRARI-BDS-2024 — van der Vlist, Helmond & Ferrari, “Big AI: Cloud infrastructure dependence and the industrialisation of artificial intelligence,” Big Data & Society, 2024
• SRC-SL-VENDROW-ET-AL-ARXIV-2502-03461 — Vendrow, Vendrow, Beery & Madry, “Do Large Language Model Benchmarks Test Reliability?,” arXiv:2502.03461, February 2025
• SRC-SL-WAN-ET-AL-FMTI-2025-ARXIV-2512-10169 — Wan, Klyman, Kapoor, Maslej, Longpre, Xiong, Liang & Bommasani, “The 2025 Foundation Model Transparency Index,” arXiv:2512.10169, December 2025
• SRC-SL-WHITFILL-WU-BECKER-RUSH-METR-2026-03-10 — Whitfill, Wu, Becker & Rush, “Many SWE-bench-Passing PRs Would Not Be Merged into Main,” METR research note, 10 March 2026
• SRC-SL-WOLFRATH-ET-AL-MODEL-RETIREMENT-ARXIV-2609-04699 — Wolfrath, Conroy, Kosten, Bell, Neupane, Kindel et al., “Model Retirement Creates Reproducibility Risk in Biomedical AI Publications,” arXiv:2609.04699, September 2026
• SRC-SL-WRIGHT-ET-AL-EPISTEMIC-DIVERSITY-ARXIV-2510-04226 — Wright, Masud, Moore et al., “What and whose knowledge? Measuring epistemic diversity in large language models,” arXiv:2510.04226, accepted to EMNLP 2026
• SRC-SL-WU-ET-AL-COUNTERFACTUAL-ARXIV-2307-02477 — Wu et al., “Reasoning or Reciting? Exploring the Capabilities and Limitations of Language Models Through Counterfactual Tasks,” NAACL 2024
• SRC-SL-XIAO-HU-XIAO-CAO-SHEN-CHI-2026 — Xiao, Hu, Xiao, Cao & Shen, “Can GenAI Move from Individual Use to Collaborative Work?,” Proceedings of CHI 2026
• SRC-SL-XUE-ET-AL-ISR-2026 — Xue, Wang, Zheng, Li & Tan, “Can ChatGPT Kill User-Generated Q&A Platforms?,” Information Systems Research, 2026
• SRC-SL-YIU-KOSOY-GOPNIK-ARXIV-2305-07666 — Yiu, Kosoy & Gopnik, “Imitation versus Innovation: What children can do that large language and language-and-vision models cannot (yet)?”, arXiv:2305.07666, 2023
• SRC-SL-YOTZOV-ET-AL-NBER-34836 — Yotzov, Barrero, Bloom et al., “Firm Data on AI,” NBER Working Paper 34836, revised March 2026
• SRC-SL-YU-FLEMING-HAMPTON-COMBEMALE-THOMPSON-ARXIV-2607-08920 — Yu, Fleming, Hampton, Combemale & Thompson, “AI Adoption in S&P 500 Firms,” arXiv:2607.08920, July 2026
• SRC-SL-ZHANG-ET-AL-GSM1K-ARXIV-2405-00332 — Zhang, Da, Lee, Robinson, Wu, Song et al., “A Careful Examination of Large Language Model Performance on Grade School Arithmetic,” NeurIPS 2024 Datasets and Benchmarks Track, arXiv:2405.00332
• SRC-SL-ZHENG-ET-AL-MT-BENCH-NEURIPS-2023 — Zheng, Chiang, Sheng et al., “Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena,” NeurIPS 2023 Datasets and Benchmarks Track, arXiv:2306.05685
• SRC-SL-ZHENG-HONG-LIU-NI-ARXIV-2607-20923 — Zheng, Hong, Liu & Ni, “Scientific exploration, collaboration and labor division in the large language model era,” arXiv:2607.20923, July 2026
• SRC-SL-ZHOU-LEE-PNAS-NEXUS-2024 — Zhou & Lee, “Generative artificial intelligence, human creativity, and art,” PNAS Nexus 3, pgae052, 2024
• SRC-SL-ZHU-ET-AL-AGENTIC-BENCHMARK-CHECKLIST-ARXIV-2507-02825 — Zhu, Jin, Pruksachatkun, Zhang, Liu, Cui et al., “Establishing Best Practices for Building Rigorous Agentic Benchmarks,” arXiv:2507.02825 (v5), 2025
• SRC-THEORY-BOWKER-STAR-1999 — Geoffrey C. Bowker & Susan Leigh Star, Sorting Things Out: Classification and Its Consequences, 1999
• SRC-THEORY-FEYERABEND-AGAINST-METHOD — Paul Feyerabend, Against Method
• SRC-THEORY-FOUCAULT-BIOPOLITICS — Michel Foucault, The Birth of Biopolitics — Collège de France lectures 1978–1979
• SRC-THEORY-FOUCAULT-STP — Michel Foucault, Security, Territory, Population — Collège de France lectures 1977–1978
• SRC-THEORY-HACKING-MOVING-TARGETS — Ian Hacking, work on kinds of people / moving targets / looping effects
• SRC-THEORY-HARDWIG-1985 — John Hardwig, “Epistemic Dependence,” The Journal of Philosophy 82(7), 1985
• SRC-THEORY-LIPSKY-STREET-LEVEL-BUREAUCRACY — Michael Lipsky, Street-Level Bureaucracy: Dilemmas of the Individual in Public Services
• SRC-THEORY-POHLE-THIEL-2020-DIGITAL-SOVEREIGNTY — Julia Pohle & Thorsten Thiel, “Digital sovereignty,” Internet Policy Review 9(4), 2020

Concepts (T01 structural prototypes)

• THEORY-CLASSIFICATION — CLASSIFICATION
• THEORY-DISCRETION — DISCRETION
• THEORY-EPISTEMIC-DEPENDENCE — EPISTEMIC DEPENDENCE
• THEORY-GOVERNMENTALITY — GOVERNMENTALITY
• THEORY-METHODOLOGICAL-PLURALISM — METHODOLOGICAL PLURALISM
• THEORY-TECHNOLOGICAL-SOVEREIGNTY — Technological sovereignty.

Radar Updates

• Radar Updates
• RADAR-LD07-PEW-SYNTHETIC-SURVEYS-2026 — RADAR-LD07-PEW-SYNTHETIC-SURVEYS-2026--v1.0

Search

• Search