🔍 Read the full analysis: Is AI Starting To Build Better AI? What The Evidence Shows on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
AI tools are contributing to research and development of AI systems, and Anthropic and OpenAI report growing automation of some work. But the available evidence does not show an AI independently designing and building its successor; a proposed U.S. ban targeting such development remains in committee.
A viral essay published November 13 has renewed debate over whether AI is beginning to build better AI, but the evidence cited so far points to systems assisting human-led research—not independently designing their own successors. Separate bills introduced in Congress would pause or restrict some advanced AI development, but neither has advanced beyond committee referral.
In an essay titled “The avalanche that hit math is coming for AI development,” podcaster Dwarkesh Patel argues that AI is already speeding up the development of AI and forecasts that the effect could grow. His essay draws on research assessments and figures published by AI companies; its projections are Patel’s analysis, not new experimental findings.
Anthropic reported that in August 2026, Claude led 26% of the company’s AI research and development work. The company said more than 90% of that work involved AI at or above a “collaborates” level, while no measured subset was fully autonomous. Anthropic describes the strongest form of recursive self-improvement as an AI system autonomously designing and developing its own successor, and says that has not been reached.
An independent evaluator, METR, assessed Anthropic’s internal data from August 12 to October 29. In a November 5 report, it estimated that Anthropic’s progress on well-specified tasks was occurring at around twice the rate seen between March 2024 and October 2025. METR attached high uncertainty to the estimate, saying the actual acceleration could be closer to 1.3 times or three times. It judged that AI assistance was likely responsible for most of the increase, while saying AI was not yet making highly significant, fully autonomous contributions to the most important parts of model training.
OpenAI has also described progress toward automating research tasks. According to The Decoder, which reviewed an OpenAI blog post that was not available for direct download, the company said it had met its goal of an “automated research intern,” based on its own measurements. OpenAI reportedly aims for a fully automated AI researcher by March 2028. The reported figures show uneven results: tasks taking under 15 minutes succeeded 86% of the time without intervention, while more than half of successful tasks estimated to take four to eight human hours required at least one human step.
Has it started? · explainer
AI already does much of the work of building AI. Not alone.
How far AI research and development is automated at one lab, on a scale from AL0 (no AI) to AL5 (no human in the loop), and where the proposed ban would draw its line.
Anthropic’s AI R&D work, August 2026
Automation levels from Epoch AI’s scale, as measured by Anthropic
Anthropic: “We are not there yet.”
- AI collaborates on most of the workAbove 90% of the work is at or above “AI collaborates”, where AI does large chunks under close human direction.
- AI leads about a quarterIn 26%, Claude completes most of a task from a high-level prompt while a human supervises.
- No closed loopNo measured part runs fully autonomously. That is the step “recursive self-improvement” usually means.
- Where would a ban start?The bill names “greatly accelerate” as a warning sign, so its reach depends on a measurement nobody yet makes in a standard way.
Sources: Anthropic, “Measurements for understanding the pace of AI development inside frontier labs” and “When AI builds itself” (figures as of August 2026, Anthropic’s own measurements); Ban Artificial Superintelligence Act of 2026, S.5493 and H.R.10538, sec. 3(6)(A). Bars show shares of Anthropic’s AI R&D work.
Graphic: Powered by Thorsten Meyer AI
Human Oversight Still Sets the Limits
The distinction between AI helping researchers and AI independently directing its own development matters for how quickly capabilities might advance and how institutions prepare for that possibility. Automation can increase the amount of work researchers complete, but the cited assessments do not establish that AI systems can set research priorities, make major training decisions, or carry a development cycle through without people.
The reported gains also come with caveats. METR’s estimate concerns progress on well-specified tasks, not a general measure of scientific discovery or a verified rate of improvement across the entire AI field. Its report says much of the most informative evidence was redacted. OpenAI’s figures, as described by The Decoder, come from the company itself and were not independently reviewed. These limitations make the direction of change relevant while leaving its size and broader implications uncertain.
The congressional proposals matter because they would regulate activity before the most autonomous version of recursive self-improvement has been demonstrated. For companies and users, however, the bills have not changed what can be developed or deployed today.
Products Worth Considering

Apple MacBook Neo 13-inch Laptop with A18 Pro Chip, 8GB Memory, 512GB SSD, Liquid Retina Display, and Touch ID – Blush

FLEXISPOT 36-Inch Height-Adjustable Standing Desk Converter with Keyboard Tray, Black

Logitech MX Master 4 for Mac Wireless Ergonomic Mouse, White Silver
As an affiliate, we earn on qualifying purchases.
Two Bills, No New Restrictions
The Ban Artificial Superintelligence Act of 2026 was introduced by Sen. Bernie Sanders as S.5493 on September 23 and by Rep. Greg Casar as H.R.10538 on September 24. The proposed legislation would pause training, modification and fine-tuning of large AI systems until a new Department of Artificial Intelligence established rules. Its scope includes recursive self-improvement and systems capable of automating or greatly accelerating AI research and development.
As of the source report’s November 16 status check, both bills had been referred to committee and had not moved further. The House bill had 12 Democratic cosponsors. A separate proposal, Rep. Ro Khanna’s “Human Control Over AI Act,” was described in a September CNBC report, but the source found no bill text to assess.
The debate returned to attention after the Associated Press projected on November 3 that Democrats would control both chambers of Congress. Supporters of the restrictions see that outcome as a possible opening in the next Congress. That is a political forecast, not an indication that either bill has passed or that current rules have changed.
““We are not there yet, and recursive self-improvement is not inevitable.””
— Anthropic, in its essay “When AI builds itself”
How Much Work AI Can Own
The evidence does not establish a single, independently verified rate at which AI is accelerating AI development. METR’s estimate covers one company and a defined category of tasks, carries high uncertainty, and relies partly on evidence Anthropic redacted. OpenAI’s reported task results are also company-generated, with no independent review cited in the account available here.
It remains unclear how much of the reported progress would carry over to less-defined research problems, how much depends on human oversight, and whether the companies’ measures can be compared directly. The source material also does not document an AI system autonomously designing and building a successor. Patel’s prediction that development could accelerate further should be read as commentary and forecasting, not confirmation that such a system exists.
Research and Legislation Ahead
The next evidence to watch is whether independent evaluators can examine more of the underlying company data and test AI contributions across a wider range of research tasks. Anthropic’s and OpenAI’s stated progress toward greater automation may provide further company-reported milestones, but those will need clear methods and external scrutiny to establish what the systems can do without human intervention.
On Capitol Hill, the Sanders and Casar bills would need to advance from committee before lawmakers could consider them further. The available status records show no such movement as of November 16, 2026. The proposed Khanna measure also cannot be assessed in detail until bill text is available. No new legal restriction is in effect based on the developments described here.
Key Questions
Has an AI system built its own successor?
The source material does not document an AI independently designing and developing its successor. Anthropic says that level of recursive self-improvement has not been reached.
What does the evidence show AI doing now?
Anthropic reports that Claude leads some of its research and development work, while METR says AI assistance likely accounts for much of the company’s faster progress on well-specified tasks. Human researchers still supervise work and make key decisions.
How certain is METR’s estimate of faster progress?
METR called its estimate highly uncertain. It put the rate at around twice the earlier pace but said it would not be surprised if the actual acceleration were closer to 1.3 times or three times.
Would the proposed U.S. ban affect AI companies now?
No. The Sanders and Casar bills had been referred to committee and had not advanced, according to the November 16 status reported in the source material. The proposals have not created new restrictions.
Is Patel’s essay new evidence that AI development is accelerating?
Patel’s November 13 essay uses existing assessments and company figures to argue that AI is speeding up AI development and to forecast further acceleration. The forecast is his analysis, not a new independent test showing autonomous AI development.
Source: ThorstenMeyerAI.com
