Ethan Mollick · Oct 20, 2026
Ethan Mollick: AI progress is uneven, and work is shifting toward agents
Markdown · JSON · Original source
Ethan Mollick argues that AI capabilities have changed substantially in the 30 months since his previous book: more models now outperform GPT-4 at its release, and systems can take on longer tasks. He says progress is uneven, with persistent weaknesses alongside major gains, and describes a shift from working alongside chatbots to delegating tasks to agents—AI systems that can carry out work with less step-by-step guidance. In his account, delegation can save time, but it also creates risks as people lose control over how agents produce results. The evidence here is Mollick’s retrospective and cited benchmarks, not proof that AI performs reliably across real-world work.
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
One Useful Thing · Article
Source details
Original source date: Oct 20, 2026 (2026-10-20T15:57:59+00:00).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-10-20T19:01:53.434Z. First observed: 2026-10-20T16:07:06.930Z. Retrieved: 2026-10-20T19:01:10.657Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
This is Mollick’s retrospective and account of his own use of AI; the evidence does not independently verify the reported results or establish reliability across real-world tasks. The METR measure described in the post ended when its researchers said their current methods could no longer measure AI task duration accurately. The larger estimate mentioned is attributed to other studies, which are not detailed here. The post promotes Mollick’s new book; its claims about AI trends and implications should be read as his perspective.
Open question
As AI agents take on longer tasks with less step-by-step guidance, how can people evaluate their work and retain meaningful oversight?
Source-stated
Mollick presents the period since his previous book as one of major AI change, including many models surpassing GPT-4 at its release.
A total of 180 AI models are now better than GPT-4 was at its release
Source-stated
He says the METR benchmark measured how much human work an AI could complete with at least a 50% chance of success from one prompt; it later became difficult to measure with the same methods.
METR said they could no longer accurately measure the amount of work an AI could do using their current methods.
Source-stated
Mollick describes AI progress as uneven: some abilities remain weak while others, such as mathematics, have improved sharply.
Some areas have stayed stubbornly weak (AI is still quite a bad long-form writer), while others, like solving math problems, went from terrible to superhuman in the same period.
Source-stated
He describes a new way of working, which he calls “the conjurer”: assigning tasks to agents and evaluating their output rather than staying involved in every step.
Conjurers summon agents when needed, assign them work, and evaluates what comes back.
Source-stated
Mollick says delegation can be powerful but brings a risk of reduced visibility into how AI produced its results.
It is the most powerful way to use AI, and it carries new risks as we lose control over what, exactly, the AI is doing to get us those results.
Source and contributor context
Ethan Mollick · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
Media was declared by the source, but no asset passed the supported availability checks. Open the original source to view it.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Ethan Mollick
Ethan Mollick is a Wharton researcher and author exploring how AI changes work, learning and decision-making.
One Useful Thing
One Useful Thing is Ethan Mollick’s publication about AI in work, education and everyday practice.
METR
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Editorial provenance
Version 1. Publication ID: b80bb4ab-b04f-4dfd-8f9f-2f46cb6d7a62.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-10-20T19:01:53.434Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: d16870cb30f3210aa24a192b6ff5768dcf21d0ee498a97caa7aff8a40e87c06f. Source text hash: 3bc28c4fd0df4d6e2e88bdb31501e0a787a11bcbec4d20367e99a3714b1d13ab.
Revision and source-check history
Version 1 · 2026-10-20T19:01:53.434Z
Ethan Mollick: AI progress is uneven, and work is shifting toward agents
Ethan Mollick argues that AI capabilities have changed substantially in the 30 months since his previous book: more models now outperform GPT-4 at its release, and systems can take on longer tasks. He says progress is uneven, with persistent weaknesses alongside major gains, and describes a shift from working alongside chatbots to delegating tasks to agents—AI systems that can carry out work with less step-by-step guidance. In his account, delegation can save time, but it also creates risks as people lose control over how agents produce results. The evidence here is Mollick’s retrospective and cited benchmarks, not proof that AI performs reliably across real-world work.
Original source
One Useful Thing · Article
Source details
Original source date: Oct 20, 2026 (2026-10-20T15:57:59+00:00).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-10-20T19:01:53.434Z. First observed: 2026-10-20T16:07:06.930Z. Retrieved: 2026-10-20T19:01:10.657Z.
Limits
This is Mollick’s retrospective and account of his own use of AI; the evidence does not independently verify the reported results or establish reliability across real-world tasks. The METR measure described in the post ended when its researchers said their current methods could no longer measure AI task duration accurately. The larger estimate mentioned is attributed to other studies, which are not detailed here. The post promotes Mollick’s new book; its claims about AI trends and implications should be read as his perspective.
Open question
As AI agents take on longer tasks with less step-by-step guidance, how can people evaluate their work and retain meaningful oversight?
Source-stated
Mollick presents the period since his previous book as one of major AI change, including many models surpassing GPT-4 at its release.
A total of 180 AI models are now better than GPT-4 was at its release
Source-stated
He says the METR benchmark measured how much human work an AI could complete with at least a 50% chance of success from one prompt; it later became difficult to measure with the same methods.
METR said they could no longer accurately measure the amount of work an AI could do using their current methods.
Source-stated
Mollick describes AI progress as uneven: some abilities remain weak while others, such as mathematics, have improved sharply.
Some areas have stayed stubbornly weak (AI is still quite a bad long-form writer), while others, like solving math problems, went from terrible to superhuman in the same period.
Source-stated
He describes a new way of working, which he calls “the conjurer”: assigning tasks to agents and evaluating their output rather than staying involved in every step.
Conjurers summon agents when needed, assign them work, and evaluates what comes back.
Source-stated
Mollick says delegation can be powerful but brings a risk of reduced visibility into how AI produced its results.
It is the most powerful way to use AI, and it carries new risks as we lose control over what, exactly, the AI is doing to get us those results.
Source and contributor context
Ethan Mollick · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
Media was declared by the source, but no asset passed the supported availability checks. Open the original source to view it.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Ethan Mollick
Ethan Mollick is a Wharton researcher and author exploring how AI changes work, learning and decision-making.
One Useful Thing
One Useful Thing is Ethan Mollick’s publication about AI in work, education and everyday practice.
METR
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Editorial provenance
Version 1. Publication ID: b80bb4ab-b04f-4dfd-8f9f-2f46cb6d7a62.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-10-20T19:01:53.434Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: d16870cb30f3210aa24a192b6ff5768dcf21d0ee498a97caa7aff8a40e87c06f. Source text hash: 3bc28c4fd0df4d6e2e88bdb31501e0a787a11bcbec4d20367e99a3714b1d13ab.
passed · 2026-10-20T19:01:53.434Z · anthropic/claude-opus-5.5
Publication: b80bb4ab-b04f-4dfd-8f9f-2f46cb6d7a62
attribution — pass: The reading attributes the argument and retrospective to Ethan Mollick, who is the post's byline author on One Useful Thing. It frames the claims as his account.
Ethan Mollick Oct 20, 2026
summary — pass: The title, summary and question match the post's themes: uneven but large AI progress, and a shift from chatbots to delegating work to agents, with oversight risks. It explains agents in plain language. It also notes that the evidence is Mollick's retrospective. The question about oversight follows from his stated risk.
AI is improving exponentially but unevenly, and the way we use it is shifting from chatbots to agents that do real work.
date — pass: The originalPublished value of 2026-10-20 matches the post's own dateline of Oct 20, 2026 and the feed entry.
Ethan Mollick Oct 20, 2026
support — pass: Each claim is directly supported by its excerpt. The summary's point that delegation saves time is grounded in Mollick's hours-to-minutes account. The limitations correctly separate METR's ended measure from the uncited larger estimate, and they note the book promotion. No unsupported benefits or costs are implied.
it did what would have taken me hours of work (all the research, all the graphing, all the layouts, all the images, all the citations…) in minutes.
media — pass: The selected media is the verified direct link to the original article, which is the most direct way to access the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.