Escape Meridian

· Nov 20, 2026

Noah Smith argues the intelligence explosion debate may be misplaced

Markdown · JSON · Original source

argues that focusing on whether an intelligence explosion is near may distract from dealing with powerful AI that could emerge through rapid progress without an explosion. He describes an intelligence explosion as a sudden leap in capability as AI improves itself. The debate, he says, turns on whether research ideas become harder to find faster than AI improves at finding them; only will tell. Smith doubts that slowing progress could prevent an explosion if one is possible: he argues it would only delay the threshold, while a complete halt is not achievable. He also questions whether an explosion would be more dangerous than rapid progress without one. In his view, AI could become superhuman at economically important tasks either way, though without an explosion the takeovers would be more staggered and irregular. Smith says attention should turn to AI already in use and the challenges he names: regulation, AI-enabled bioterror, widely sharing economic benefits, and competition policy. His claims about future capabilities and economic effects remain expectations, not demonstrated outcomes; he also says the evidence for some arguments is limited.

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.

Source image for Does it really matter if there’s an intelligence explosion?
Noahpinion · Open original source

Original source

Noahpinion · Article

Source details

Original source date: Nov 20, 2026 (2026-11-20T10:13:06+00:00).

Date basis: Date stated by the publisher.

Added to Escape Meridian: 2026-11-21T02:16:37.964Z. First observed: 2026-11-20T10:16:09.389Z. Retrieved: 2026-11-21T02:15:51.979Z.

Referenced by

Recorded citations in public readings. A reference alone does not establish agreement or influence.

No public references recorded here yet.

Reference data (JSON)

Limits

Smith notes that the report relies on data that aren’t public, so assessing it requires guesswork and assumptions. Smith says the economic model estimating RSI is based on only 16 data points and that he does not trust it. Smith's claims about future AI capabilities and the relative economic effects of explosive and non-explosive progress are arguments and expectations, not demonstrated outcomes.

Open question

What evidence would show whether research ideas are becoming harder to find faster than AI is improving at finding them?

Source-stated

Smith argues that debate about an intelligence explosion may distract from addressing AI already in use and expected to arrive soon.

This is why I think this whole debate is misplaced. We need to be thinking about how to deal with the AI we already have

Source-stated

Smith says the debate depends on whether AI's ability to find research ideas will improve faster than ideas become harder to find.

Basically, it depends on whether ideas get harder to find faster than AI gets better at finding ideas, and only time will tell which one wins.

Source-stated

Smith argues that slowing AI progress would delay, but not prevent, an intelligence explosion if one is possible; he doubts a complete halt can be achieved.

If there’s some threshold at which AI gets good enough to start an intelligence explosion, that threshold will be reached.

Source-stated

Smith argues that AI could become superhuman in economically important tasks without an intelligence explosion, with takeovers more staggered and irregular.

Whereas if an intelligence explosion doesn’t occur, these takeovers are likely to be staggered and irregular. But they will still happen

Source-stated

Smith calls for attention to regulation, AI-enabled bioterror, sharing AI's economic benefits, and competition policy.

We need to think about how to regulate that actually existing superintelligence, how to prevent AI-enabled bioterror, how to make sure the economic benefits of AI are widely shared

Source and contributor context

Noah Smith · Article author

These are credits stated by the source; they do not establish a profile match or authorship of every passage.

Source media

Media declared by the original publisher; availability checked separately from the editorial reading.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

Noah Smith

Economic interpretation can separate technological usefulness from financing and deployment dynamics.

TIME

TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.

METR

METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.

Editorial provenance

Version 2. Publication ID: 439883e6-9356-4ca7-8b5d-5731c396c9d4.

Prepared by gpt-6-luna; grounding review by gpt-6-luna.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-21T02:16:37.964Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: useful_perspective.

Observation: c8c65d5afc5ed76b40efd30718c2821218ab0ade203d6267d3cef16d4526d8bc. Source text hash: 6ced502b22362c8bbe0adc03eb14ab09ded11fb6020a5714cf1985fae75ec7f8.

Correction reason: Revised the summary to clarify Smith’s central argument, preserve his stated uncertainty, and note the limited evidence he identifies.

Revision and source-check history

Version 2 · 2026-11-21T02:16:37.964Z · Revised the summary to clarify Smith’s central argument, preserve his stated uncertainty, and note the limited evidence he identifies.

Noah Smith argues the intelligence explosion debate may be misplaced

argues that focusing on whether an intelligence explosion is near may distract from dealing with powerful AI that could emerge through rapid progress without an explosion. He describes an intelligence explosion as a sudden leap in capability as AI improves itself. The debate, he says, turns on whether research ideas become harder to find faster than AI improves at finding them; only will tell. Smith doubts that slowing progress could prevent an explosion if one is possible: he argues it would only delay the threshold, while a complete halt is not achievable. He also questions whether an explosion would be more dangerous than rapid progress without one. In his view, AI could become superhuman at economically important tasks either way, though without an explosion the takeovers would be more staggered and irregular. Smith says attention should turn to AI already in use and the challenges he names: regulation, AI-enabled bioterror, widely sharing economic benefits, and competition policy. His claims about future capabilities and economic effects remain expectations, not demonstrated outcomes; he also says the evidence for some arguments is limited.

Original source

Noahpinion · Article

Source details

Original source date: Nov 20, 2026 (2026-11-20T10:13:06+00:00).

Date basis: Date stated by the publisher.

Added to Escape Meridian: 2026-11-21T02:16:37.964Z. First observed: 2026-11-20T10:16:09.389Z. Retrieved: 2026-11-21T02:15:51.979Z.

Limits

Smith notes that the report relies on data that aren’t public, so assessing it requires guesswork and assumptions. Smith says the economic model estimating RSI is based on only 16 data points and that he does not trust it. Smith's claims about future AI capabilities and the relative economic effects of explosive and non-explosive progress are arguments and expectations, not demonstrated outcomes.

Open question

What evidence would show whether research ideas are becoming harder to find faster than AI is improving at finding them?

Source-stated

Smith argues that debate about an intelligence explosion may distract from addressing AI already in use and expected to arrive soon.

This is why I think this whole debate is misplaced. We need to be thinking about how to deal with the AI we already have

Source-stated

Smith says the debate depends on whether AI's ability to find research ideas will improve faster than ideas become harder to find.

Basically, it depends on whether ideas get harder to find faster than AI gets better at finding ideas, and only time will tell which one wins.

Source-stated

Smith argues that slowing AI progress would delay, but not prevent, an intelligence explosion if one is possible; he doubts a complete halt can be achieved.

If there’s some threshold at which AI gets good enough to start an intelligence explosion, that threshold will be reached.

Source-stated

Smith argues that AI could become superhuman in economically important tasks without an intelligence explosion, with takeovers more staggered and irregular.

Whereas if an intelligence explosion doesn’t occur, these takeovers are likely to be staggered and irregular. But they will still happen

Source-stated

Smith calls for attention to regulation, AI-enabled bioterror, sharing AI's economic benefits, and competition policy.

We need to think about how to regulate that actually existing superintelligence, how to prevent AI-enabled bioterror, how to make sure the economic benefits of AI are widely shared

Source and contributor context

Noah Smith · Article author

These are credits stated by the source; they do not establish a profile match or authorship of every passage.

Source media

Media declared by the original publisher; availability checked separately from the editorial reading.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

Noah Smith

Economic interpretation can separate technological usefulness from financing and deployment dynamics.

TIME

TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.

METR

METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.

Editorial provenance

Version 2. Publication ID: 439883e6-9356-4ca7-8b5d-5731c396c9d4.

Prepared by gpt-6-luna; grounding review by gpt-6-luna.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-21T02:16:37.964Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: useful_perspective.

Observation: c8c65d5afc5ed76b40efd30718c2821218ab0ade203d6267d3cef16d4526d8bc. Source text hash: 6ced502b22362c8bbe0adc03eb14ab09ded11fb6020a5714cf1985fae75ec7f8.

Correction reason: Revised the summary to clarify Smith’s central argument, preserve his stated uncertainty, and note the limited evidence he identifies.

Version 1 · 2026-11-20T10:17:28.004Z

Noah Smith: Does it really matter if there’s an intelligence explosion?

argues that debate over whether an intelligence explosion will happen may distract from dealing with AI already in use and the more powerful AI he expects soon. He describes an intelligence explosion as a sudden leap in capability driven by AI improving itself. The debate, he says, turns on whether ideas in AI research get harder to find faster than AI gets better at finding them; only will tell. Smith doubts that slowing progress would prevent an explosion: if one is possible, he argues, slowing progress would only delay it, while a complete halt is not achievable. He also questions whether an explosion would be more dangerous than rapid progress without one. In his view, AI could become superhuman at economically important tasks either way, though without an explosion its takeovers would be more staggered and irregular. Smith urges attention to the AI already in use and the challenges he names: regulation, AI-enabled bioterror, widely sharing economic benefits, and competition policy. These are his arguments and expectations, not established outcomes; he acknowledges uncertainty about whether an explosion will occur and how its economic effects would compare.

Original source

Noahpinion · Article

Source details

Original source date: Nov 20, 2026 (2026-11-20T10:13:06+00:00).

Date basis: Date stated by the publisher.

Added to Escape Meridian: 2026-11-20T10:17:28.004Z. First observed: 2026-11-20T10:16:09.389Z. Retrieved: 2026-11-20T10:16:34.968Z.

Limits

Smith notes that the report relies on data that aren’t public, so assessing it requires guesswork and assumptions. Smith says the economic model estimating RSI is based on only 16 data points and that he does not trust it. Smith's claims about future AI capabilities and the relative economic effects of explosive and non-explosive progress are arguments and expectations, not demonstrated outcomes.

Open question

Will ideas in AI research become harder to find faster than AI improves at finding them, as the competing arguments Smith describes suggest?

Source-stated

Smith argues that debate about an intelligence explosion may distract from addressing AI already in use and expected to arrive soon.

This is why I think this whole debate is misplaced. We need to be thinking about how to deal with the AI we already have

Source-stated

Smith says the debate depends on whether AI's ability to find research ideas will improve faster than ideas become harder to find.

Basically, it depends on whether ideas get harder to find faster than AI gets better at finding ideas, and only time will tell which one wins.

Source-stated

Smith argues that slowing AI progress would delay, but not prevent, an intelligence explosion if one is possible; he doubts a complete halt can be achieved.

If there’s some threshold at which AI gets good enough to start an intelligence explosion, that threshold will be reached.

Source-stated

Smith argues that AI could become superhuman in economically important tasks without an intelligence explosion, with takeovers more staggered and irregular.

Whereas if an intelligence explosion doesn’t occur, these takeovers are likely to be staggered and irregular. But they will still happen

Source-stated

Smith calls for attention to regulation, AI-enabled bioterror, sharing AI's economic benefits, and competition policy.

We need to think about how to regulate that actually existing superintelligence, how to prevent AI-enabled bioterror, how to make sure the economic benefits of AI are widely shared

Source and contributor context

Noah Smith · Article author

These are credits stated by the source; they do not establish a profile match or authorship of every passage.

Source media

Media declared by the original publisher; availability checked separately from the editorial reading.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

Noah Smith

Economic interpretation can separate technological usefulness from financing and deployment dynamics.

TIME

TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.

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: 5af554d3-58f8-43ac-8852-9bffc51398e1.

Prepared by gpt-6-luna; grounding review by gpt-6-luna.

Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-20T10:17:28.004Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: useful_perspective.

Observation: d640ef36eeb1aab7949bdfb2e91caf446765d9ecb24b62eab2ed3e92917b7e7f. Source text hash: 3311aa6d9bbc75d81efe168e07f97ae27166fa602e28d75b5a7ed177af11ca22.

uncertain · 2026-11-20T10:17:28.004Z · anthropic/claude-opus-5.5

Publication: 5af554d3-58f8-43ac-8852-9bffc51398e1

attribution — pass: The reading consistently attributes the arguments to Noah Smith, the author and publisher of the Noahpinion post. It also keeps the METR report and the 16-data-point model caveats attributed to Smith's own commentary.

Noah Smith Nov 20, 2026

summary — uncertain: The core argument is conveyed accessibly, and intelligence explosion is defined. Three repairs would restore fidelity. The question ends 'as the competing arguments Smith describes suggest', which implies those arguments point to one answer; they conflict. Fix: drop that clause or say 'the question Smith says decides the debate'. 'May distract' softens Smith's flat verdict that the debate is 'misplaced'. Fix: use 'is misplaced'. The alignment point (alignment research would be left behind in an explosion) is erased. Fix: add one clause.

This is why I think this whole debate is misplaced.

date — pass: The originalPublished value of 2026-11-20 matches the byline date on the publisher page and the feed entry.

Noah Smith Nov 20, 2026

support — pass: Each claim matches its quoted excerpt and is framed as Smith's argument. The limitations accurately reflect his caveats about the non-public METR data and the 16 data points. The summary's 'could become superhuman' is softer than Smith's 'will', but it does not add an unsupported claim. No invented benefits or costs appear.

This estimation is done with only 16 data points, so I don’t really trust it.

media — pass: The selected media is the verified direct link to the original article, which is the most direct way to access the source.

passed · 2026-11-21T02:16:37.964Z · anthropic/claude-opus-5.5

Publication: 439883e6-9356-4ca7-8b5d-5731c396c9d4

attribution — pass: The reading consistently attributes the arguments to Noah Smith, the author and publisher of the Noahpinion post. Its limitations attribute the METR data caveat and the 16-data-point caveat to Smith's own commentary, which matches the source.

This is why I think this whole debate is misplaced.

summary — pass: The title, summary and question match Smith's thesis: the intelligence explosion debate is misplaced; slowing progress only delays a threshold; a complete halt is impossible; non-explosive progress may look similar; and attention should turn to present and near-term AI. The term is explained plainly. Hedges like 'may' and 'could' soften Smith's confident 'will' somewhat, and his alignment point is compressed, but his meaning stays intact. The question follows from the crux he names.

It’s not possible to stop AI improvement completely, and so if an intelligence explosion is possible, it’s going to happen to us soon.

date — pass: originalPublished of 2026-11-20 matches the article byline date and the feed entry. No earlier publication is indicated.

Noah Smith Nov 20, 2026

support — pass: Each claim's excerpt appears verbatim in the source and supports the paraphrase. The limitations about non-public METR data and the 16-data-point estimate are source-stated. The draft correctly frames forecasts as Smith's expectations rather than demonstrated outcomes, and it adds no unsupported benefits or costs.

This estimation is done with only 16 data points, so I don’t really trust it.

media — pass: The selected media is the verified direct link to the original Noahpinion article, which is the most direct way to reach 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.