You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/search.json
+1-29Lines changed: 1 addition & 29 deletions
Original file line number
Diff line number
Diff line change
@@ -4,7 +4,7 @@
4
4
"href": "index.html",
5
5
"title": "Blog Posts",
6
6
"section": "",
7
-
"text": "Formal derivations\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nLLM verification\n\n\n\n\n\n\n\n\n\n\nFeb 3, 2026\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nForecasts of AI & Economic Growth\n\n\n\n\n\n\n\n\n\n\nNov 6, 2025\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nEconomics and Transformative AI\n\n\n\n\n\n\n\n\n\n\nOct 2, 2025\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nOn Deriving Things\n\n\n\n\n\n\n\n\n\n\nJan 30, 2025\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nToo Much Good News is Bad News\n\n\n\n\n\n\n\n\n\n\nDec 26, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nPremature Optimization and the Valley of Confusion\n\n\n\n\n\n\n\n\n\n\nMay 10, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nPeer Effects, Culture, and Taxes\n\n\n\n\n\n\n\n\n\n\nApr 28, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nBloodhounds and Bulldogs\n\n\nOn Perception, Judgment, & Decision-Making\n\n\n\n\n\n\n\nApr 27, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nThe Influence of AI on Content Moderation and Communication\n\n\n\n\n\n\n\n\n\n\nDec 11, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nThe History of Automated Text Moderation\n\n\n\n\n\n\n\n\n\n\nNov 18, 2023\n\n\nIntegrity Institute collaborators: Alex Rosenblatt, Jeff Allen, Ejona Varangu, Dave Sullivan, Tom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nThinking About Tradeoffs? Draw an Ellipse\n\n\n\n\n\n\n\n\n\n\nOct 25, 2023\n\n\nTom Cunningham, OpenAI.\n\n\n\n\n\n\n\n\n\n\n\n\nExperiment Interpretation and Extrapolation\n\n\n\n\n\n\n\n\n\n\nOct 17, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nAn AI Which Imitates Humans Can Beat Humans\n\n\n\n\n\n\n\n\n\n\nOct 6, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nSushi-Roll Model of Online Media\n\n\nPreviously: “pizza model”, “salami model”\n\n\n\n\n\n\n\nSep 8, 2023\n\n\nTom Cunningham, Integrity Institute\n\n\n\n\n\n\n\n\n\n\n\n\nHow Much has Social Media affected Polarization?\n\n\n\n\n\n\n\n\n\n\nAug 7, 2023\n\n\nTom Cunningham, Integrity Institute\n\n\n\n\n\n\n\n\n\n\n\n\nThe Paradox of Small Effects\n\n\n\n\n\n\n\n\n\n\nAug 2, 2023\n\n\nTom Cunningham, Integrity Institute\n\n\n\n\n\n\n\n\n\n\n\n\nRanking by Engagement\n\n\n\n\n\n\n\n\n\n\nMay 8, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nSocial Media Suspensions of Prominent Accounts\n\n\n\n\n\n\n\n\n\n\nJan 31, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nOptimal Coronavirus Policy Should be Front-Loaded\n\n\n\n\n\n\n\n\n\n\nApr 5, 2020\n\n\n\n\n\n\n\n\n\n\n\n\nOn Unconscious Influences (Part 1)\n\n\n\n\n\n\n\n\n\n\nDec 8, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nThe Work of Art in the Age of Mechanical Production\n\n\n\n\n\n\n\n\n\n\nSep 27, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nRepulsion from the Prior\n\n\n\n\n\n\n\n\n\n\nMay 26, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nThe Repeated Failure of Laws of Behaviour\n\n\n\n\n\n\n\n\n\n\nApr 15, 2017\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nEconomist Explorers\n\n\n\n\n\n\n\n\n\n\nFeb 25, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nSamuelson & Expected Utility\n\n\n\n\n\n\n\n\n\n\nFeb 25, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nWeber’s Law Doesn’t Imply Concave Representations or Concave Judgments\n\n\n\n\n\n\n\n\n\n\nFeb 25, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nRelative Thinking\n\n\n\n\n\n\n\n\n\n\nApr 30, 2016\n\n\nTom Cunningham\n\n\n\n\n\nNo matching items"
7
+
"text": "LLM verification\n\n\n\n\n\n\n\n\n\n\nFeb 5, 2026\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nForecasts of AI & Economic Growth\n\n\n\n\n\n\n\n\n\n\nNov 6, 2025\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nEconomics and Transformative AI\n\n\n\n\n\n\n\n\n\n\nOct 2, 2025\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nOn Deriving Things\n\n\n\n\n\n\n\n\n\n\nJan 30, 2025\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nToo Much Good News is Bad News\n\n\n\n\n\n\n\n\n\n\nDec 26, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nPremature Optimization and the Valley of Confusion\n\n\n\n\n\n\n\n\n\n\nMay 10, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nPeer Effects, Culture, and Taxes\n\n\n\n\n\n\n\n\n\n\nApr 28, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nBloodhounds and Bulldogs\n\n\nOn Perception, Judgment, & Decision-Making\n\n\n\n\n\n\n\nApr 27, 2024\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nThe Influence of AI on Content Moderation and Communication\n\n\n\n\n\n\n\n\n\n\nDec 11, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nThe History of Automated Text Moderation\n\n\n\n\n\n\n\n\n\n\nNov 18, 2023\n\n\nIntegrity Institute collaborators: Alex Rosenblatt, Jeff Allen, Ejona Varangu, Dave Sullivan, Tom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nThinking About Tradeoffs? Draw an Ellipse\n\n\n\n\n\n\n\n\n\n\nOct 25, 2023\n\n\nTom Cunningham, OpenAI.\n\n\n\n\n\n\n\n\n\n\n\n\nExperiment Interpretation and Extrapolation\n\n\n\n\n\n\n\n\n\n\nOct 17, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nAn AI Which Imitates Humans Can Beat Humans\n\n\n\n\n\n\n\n\n\n\nOct 6, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nSushi-Roll Model of Online Media\n\n\nPreviously: “pizza model”, “salami model”\n\n\n\n\n\n\n\nSep 8, 2023\n\n\nTom Cunningham, Integrity Institute\n\n\n\n\n\n\n\n\n\n\n\n\nHow Much has Social Media affected Polarization?\n\n\n\n\n\n\n\n\n\n\nAug 7, 2023\n\n\nTom Cunningham, Integrity Institute\n\n\n\n\n\n\n\n\n\n\n\n\nThe Paradox of Small Effects\n\n\n\n\n\n\n\n\n\n\nAug 2, 2023\n\n\nTom Cunningham, Integrity Institute\n\n\n\n\n\n\n\n\n\n\n\n\nRanking by Engagement\n\n\n\n\n\n\n\n\n\n\nMay 8, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nSocial Media Suspensions of Prominent Accounts\n\n\n\n\n\n\n\n\n\n\nJan 31, 2023\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nOptimal Coronavirus Policy Should be Front-Loaded\n\n\n\n\n\n\n\n\n\n\nApr 5, 2020\n\n\n\n\n\n\n\n\n\n\n\n\nOn Unconscious Influences (Part 1)\n\n\n\n\n\n\n\n\n\n\nDec 8, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nThe Work of Art in the Age of Mechanical Production\n\n\n\n\n\n\n\n\n\n\nSep 27, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nRepulsion from the Prior\n\n\n\n\n\n\n\n\n\n\nMay 26, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nThe Repeated Failure of Laws of Behaviour\n\n\n\n\n\n\n\n\n\n\nApr 15, 2017\n\n\nTom Cunningham\n\n\n\n\n\n\n\n\n\n\n\n\nEconomist Explorers\n\n\n\n\n\n\n\n\n\n\nFeb 25, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nSamuelson & Expected Utility\n\n\n\n\n\n\n\n\n\n\nFeb 25, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nWeber’s Law Doesn’t Imply Concave Representations or Concave Judgments\n\n\n\n\n\n\n\n\n\n\nFeb 25, 2017\n\n\n\n\n\n\n\n\n\n\n\n\nRelative Thinking\n\n\n\n\n\n\n\n\n\n\nApr 30, 2016\n\n\nTom Cunningham\n\n\n\n\n\nNo matching items"
"title": "Optimal Coronavirus Policy Should be Front-Loaded",
972
972
"section": "Prior Discussion",
973
973
"text": "Prior Discussion\nThere’s been some discussion of zig-zagging by the Imperial group (paper) and by Timothy Gowers (twitter & post)\nGowers says the optimal policy is very short zig-zags (changing policy every other day), however I think this is misleading. It comes from fixing the lower-threshold and optimizing the upper-threshold. If instead you fixed the upper-threshold and optimized the lower-threshold, then the optimal cycle-length will be long.\nIf you choose both the upper and lower threshold (both T and S) then he notes that they’ll both be arbitarily low. However this ignores the cost of getting to zero given current cases.\nInstead a well-defined problem is to choose an optimal time-path of policy given some start-point and end-point. In that case it’ll be a path of gradually decreasing strictness (without zig-zags).\nYou can see the intuition in the diagram below: the total infections is approximately the area under the zig-zag (not quite: because the y-axis is ln(cases), but this won’t matter for the argument). Thus you can reduce the area under the line by lowering the upper threshold. However if you instead take the upper threshold as fixed, then it’s optimal to choose a lower threshold that is as low as possible, i.e. you want long cycles, not short cycles.\n\n\n\nabc"
"text": "Model of LLMs for Discovery\n\nKnowledge-sharing LLMs.\n\nTraditionally LLMs have been trained with human judgment as the ground truth, using labels from paid raters or from customers. As a consequence they can answer questions and solve problems up to the limits of human expertise but rarely beyond (with some exceptions, see the literature on LLM “Transcendence”, Abreu et al. (2025)).\nIf we model the economic effects of LLMs as coming from sharing existing knowledge this has a number of implications that seem to fit the data.1 (See also some of my previous writing on this: AI & Imitation, a pocket model of AI).\n\nLLMs use will be higher among those junior in their careers, facing problems that are new to them.\nLLMs will be dispropportionately used by people outside their area of expertise, e.g. lawyers will use them for medical questions, doctors will use them for legal questions.\nLLMs will be disproportionately used in well-documented domains, e.g. for popular programming languages, which appear more in the training data.\nLLMs will decrease knowledge rents – the premium earned by people and firms whose value comes from knowledge.\nLLMs will increase home production – you can solve problems yourself instead of paying for it, and so potentially decrease GDP.\nLLMs decrease the returns to innovation and news-gathering, because they increase the speed of knowledge diffusion and thus diminish the rents that can earned from new knowledge.\nLLM use has high fixed costs (collecting the knowledge) and low marginal costs in sharing the knowledge. The returns to LLM-use on an individual problem rapidly diminish when you hit the frontier of existing knowledge.\n\n\n\n\nKnowledge-creating LLMs.\n\nOver the past 18 months it has become much more popular to train LLMs directly against a source of ground truth, e.g. Reinforcement Learning against Verifiable Rewards (RLVR). Accompanying this there has been a steadily increasing stream of announcements of new discoveries by LLMs.\nKnowledge-creating LLMs are distinct from prior AI discovery applications (e.g. AlphaFold, AlphaTensor) in that they are general methods. E.g. Yuksekgonul et al. (2026) describes a setup in which a general-purpose LLM iteratively explores any arbitrary optimization landscape, using a combination of weight-updates and an explicit scratchpad state.\nKnowledge-creating LLMs will differ from knowledge-sharing LLMs in a number of ways:\n\nKnowledge-creating LLMs will have qualitatively different benchmarks: instead of seeing if they can answer questions which we already know the answer to (most existing benchmarks), we want them to answer new questions, e.g. solve an Erdős problem or set a new record on an optimization problem.\nKnowledge-creating LLMs have high returns to compute expenditure on individual problems, unlike knowledge-sharing LLMs for which returns asymptote quickly.\nKnowledge-creating LLMs will be adopted by leader firms more than followers.\nThe demand for new knowledge is much less elastic than the demand for existing knowledge, i.e. there are high returns to exclusivity of new knowledge. Thus LLM-providers are likely to license their technology exclusively rather than expose them through a general-purpose API.\n\nSarah Friar, OpenAI’s CFO, January 2026.\n\n“As intelligence moves into scientific research, drug discovery, energy systems, and financial modeling, new economic models will emerge. Licensing, IP-based agreements, and outcome-based pricing will share in the value created.”\n\n\nApplications for knowledge-creating LLMs:\n\n\nDrug discovery.\nOptimizing algorithms.\nAI R&D.\nPredict stock prices."
"text": "Footnotes\n\n\nMany other technologies share knowledge – speaking, writing, printing, the internet – LLMs just continue this progression but further lower the costs of sharing.↩︎"
0 commit comments