# Marginal Revolution: AI’s gains and risks in a weak-links economy [Canonical reading](https://escapemeridian.com/readings/marginal-revolution-ai-s-gains-and-risks-in-a-weak-links-economy-ce154450) Marginal Revolution relays an argument from an in-progress manuscript by Charles I. Jones and Christopher Tonetti: in a “weak links” model, output is constrained by the least productive tasks. Because tasks are complements, automation can raise productivity without lifting output much, while attacks that sharply reduce productivity in targeted tasks can cause much greater harm. The post gives a numerical example: automating tasks worth 1 percent of GDP with a tenfold productivity increase raises output by less than one percent, while cutting their productivity tenfold lowers output by 68 percent on impact. It argues that if a larger AI sector both improves AI models and makes attacks more likely, the productivity spillover may be negligible beside the potential damage, leading the market to overinvest in AI. The authors’ central point, as relayed by the post, is that AI’s costs and benefits can be extremely asymmetric in a weak-link economy. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source Marginal Revolution · Article Original source date: Nov 15, 2026 (2026-11-15T10:03:58+00:00). Date basis: Date stated by the publisher. Added to Escape Meridian: 2026-11-15T10:16:46.806Z. First observed: 2026-11-15T10:06:20.568Z. Retrieved: 2026-11-15T10:15:48.060Z. - [AI and complements - Marginal REVOLUTION](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Limits The post relays an argument from an in-progress manuscript and does not provide further details about the attacks or the basis for the claim that a larger AI sector makes them more likely. ## Open question What in the manuscript supports the claim that a larger AI sector makes attacks more likely? ## Source-stated The post describes a “weak links” model in which output is constrained by the least productive tasks. > output is constrained by the least productive tasks. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source-stated In the example, automation produces a smaller output gain than the loss caused by a tenfold productivity cut to the same tasks. > raises output by less than one percent, while cutting their productivity tenfold lowers output by 68 percent on impact. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source-stated The post reports the manuscript’s argument that a larger AI sector could improve AI models while making attacks more likely, so potential damage may outweigh the productivity spillover and lead to market overinvestment. > the positive productivity spillover is negligible compared with the potential damage, so the market overinvests in AI. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source-stated The post characterizes the costs and benefits of AI in a weak-link economy as extremely asymmetric. > the costs and benefits from AI can be extremely asymmetric. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source and contributor context Tyler Cowen · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Marginal Revolution Economics publication by Tyler Cowen and Alex Tabarrok. - [Explore this context](/entities/marginal-revolution) - [Reference](https://marginalrevolution.com/) ## Tyler Cowen Economics, institutions and unexpected cross-domain questions; useful source discovery. - [Explore this context](/entities/tyler-cowen) - [Reference](https://www.dwarkesh.com/p/tyler-cowen-4) - [Reference](https://www.dwarkesh.com/p/tyler-cowen-3) - [Reference](https://www.dwarkesh.com/p/tyler-cowen-2) - [Reference](https://marginalrevolution.com/marginalrevolution/2026/10/saturday-assorted-links-583.html?utm_source=rss&utm_medium=rss&utm_campaign=saturday-assorted-links-583) - [Reference](https://tylercowen.com/) ## Editorial provenance Version 1. Publication ID: 9871055b-6ead-4bbc-8995-d80b86ca1edc. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-15T10:16:46.806Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: useful\_perspective. Observation: 17ac9e4f803f0b35bfc72fd2c00c88bb33ab558a11457cc81a8a44317c063b94. Source text hash: b3b96eaa34776cdfce315effd75c554b1828c0285cd80dcc9d1fe3e92d2ec17e. ## Revision and source-check history ### Version 1 — 2026-11-15T10:16:46.806Z Marginal Revolution relays an argument from an in-progress manuscript by Charles I. Jones and Christopher Tonetti: in a “weak links” model, output is constrained by the least productive tasks. Because tasks are complements, automation can raise productivity without lifting output much, while attacks that sharply reduce productivity in targeted tasks can cause much greater harm. The post gives a numerical example: automating tasks worth 1 percent of GDP with a tenfold productivity increase raises output by less than one percent, while cutting their productivity tenfold lowers output by 68 percent on impact. It argues that if a larger AI sector both improves AI models and makes attacks more likely, the productivity spillover may be negligible beside the potential damage, leading the market to overinvest in AI. The authors’ central point, as relayed by the post, is that AI’s costs and benefits can be extremely asymmetric in a weak-link economy. ## Original source Marginal Revolution · Article Original source date: Nov 15, 2026 (2026-11-15T10:03:58+00:00). Date basis: Date stated by the publisher. Added to Escape Meridian: 2026-11-15T10:16:46.806Z. First observed: 2026-11-15T10:06:20.568Z. Retrieved: 2026-11-15T10:15:48.060Z. - [AI and complements - Marginal REVOLUTION](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Limits The post relays an argument from an in-progress manuscript and does not provide further details about the attacks or the basis for the claim that a larger AI sector makes them more likely. ## Open question What in the manuscript supports the claim that a larger AI sector makes attacks more likely? ## Source-stated The post describes a “weak links” model in which output is constrained by the least productive tasks. > output is constrained by the least productive tasks. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source-stated In the example, automation produces a smaller output gain than the loss caused by a tenfold productivity cut to the same tasks. > raises output by less than one percent, while cutting their productivity tenfold lowers output by 68 percent on impact. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source-stated The post reports the manuscript’s argument that a larger AI sector could improve AI models while making attacks more likely, so potential damage may outweigh the productivity spillover and lead to market overinvestment. > the positive productivity spillover is negligible compared with the potential damage, so the market overinvests in AI. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source-stated The post characterizes the costs and benefits of AI in a weak-link economy as extremely asymmetric. > the costs and benefits from AI can be extremely asymmetric. - [Supporting source](https://marginalrevolution.com/marginalrevolution/2026/11/ai-and-complements.html) ## Source and contributor context Tyler Cowen · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Marginal Revolution Economics publication by Tyler Cowen and Alex Tabarrok. - [Explore this context](/entities/marginal-revolution) - [Reference](https://marginalrevolution.com/) ## Tyler Cowen Economics, institutions and unexpected cross-domain questions; useful source discovery. - [Explore this context](/entities/tyler-cowen) - [Reference](https://www.dwarkesh.com/p/tyler-cowen-4) - [Reference](https://www.dwarkesh.com/p/tyler-cowen-3) - [Reference](https://www.dwarkesh.com/p/tyler-cowen-2) - [Reference](https://marginalrevolution.com/marginalrevolution/2026/10/saturday-assorted-links-583.html?utm_source=rss&utm_medium=rss&utm_campaign=saturday-assorted-links-583) - [Reference](https://tylercowen.com/) ## Editorial provenance Version 1. Publication ID: 9871055b-6ead-4bbc-8995-d80b86ca1edc. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-15T10:16:46.806Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: useful\_perspective. Observation: 17ac9e4f803f0b35bfc72fd2c00c88bb33ab558a11457cc81a8a44317c063b94. Source text hash: b3b96eaa34776cdfce315effd75c554b1828c0285cd80dcc9d1fe3e92d2ec17e. ### uncertain — 2026-11-15T10:16:46.806Z — anthropic/claude-opus-5.5 Publication: 9871055b-6ead-4bbc-8995-d80b86ca1edc attribution — pass: The draft correctly presents Marginal Revolution as relaying an in-progress manuscript by Charles I. Jones and Christopher Tonetti, via Lukas Freund. Phrases such as 'The post gives a numerical example' and 'It argues' refer to the quoted manuscript text, but the opening sentence and the final 'as relayed by the post' keep authorship clear. > That is from a recent “in progress” manuscript by Charles I. Jones and Christopher Tonetti. Via Lukas Freund. summary — uncertain: The numerical example drops its key condition. The 68 percent loss and the gain of under 1 percent hold only with an elasticity of substitution across tasks of 0.2, which measures how hard it is to substitute one task for another. Without it, the figures read as general results. The draft also omits that complementarity makes attacks 'easier to execute,' and it never explains 'complements' in plain language. Smallest repair: add 'assuming an elasticity of substitution across tasks of 0.2 (tasks are hard to substitute for one another).' Also restore 'and easier to execute.' > For example, with an elasticity of substitution across tasks of 0.2, automating tasks worth 1 percent of GDP date — pass: The originalPublished value, 2026-11-15T10:03:58+00:00 from the publisher page, matches the feed entry and the first observation. This is the post's own publication date. The manuscript's date is not given, and the draft does not claim one. support — pass: Each claim is source-stated with an exact excerpt. The overinvestment claim keeps the source's conditional framing ('When a larger AI sector simultaneously...'). The limitation correctly notes that the post gives no basis for the attack-likelihood premise. The question follows from that gap and is not leading. No unsupported benefits or costs are introduced. > Whena larger AI sector simultaneously makes AI models more capable and attacks more likely media — pass: The selected media is the verified, direct link to the original Marginal Revolution post. No other media was found. It is the most direct access to the source. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/ce1544502244033c6ace934408f064075e5dd3ed43d3fccaa13d32360c951013/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/marginal-revolution-ai-s-gains-and-risks-in-a-weak-links-economy-ce154450.json)