📊 Full opportunity report: When AI Builds Itself: Inside Anthropic’s Evidence on Recursive Self-Improvement on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Anthropic’s new report provides data indicating AI systems are already automating parts of AI research and development. While human decision-making remains a bottleneck, the evidence suggests rapid progress toward autonomous AI improvement, though full recursive self-improvement is not yet achieved.
Anthropic’s recent analysis presents measurable evidence that AI systems are increasingly automating aspects of AI research and development, with the potential to reach recursive self-improvement if certain human-guided decision points are automated. This development could accelerate AI progress significantly, though experts emphasize that full self-improvement remains a future possibility, not an imminent reality.
The report from The Anthropic Institute highlights that AI models like Claude are now capable of performing many tasks involved in AI development, including writing code and conducting experiments, with a marked increase in productivity. For example, Anthropic engineers now ship eight times as much code per quarter as they did in 2021–2025, indicating a rapid acceleration in AI-assisted development.
Public benchmarks such as METR show that AI’s ability to handle complex tasks has doubled roughly every four months, with models now capable of managing tasks that previously required days, within hours or a few days. Internal data from Anthropic further reveals that over 80% of code merged into their projects is now authored by AI, a significant jump from early 2025 when it was in the low single digits.
Despite these advances, the report emphasizes that the critical bottleneck—human decision-making in choosing goals and evaluating results—remains largely outside AI control. The authors suggest that if this decision-making process can also be automated, the cycle of self-improvement could accelerate dramatically, potentially running at the speed of compute rather than human effort.
When AI builds itself
Anthropic is delegating a growing share of AI development to AI. Taken far enough, that points to a system that designs its own successor — recursive self-improvement. Not here yet, not inevitable. But the case isn’t speculation: it’s data on what AI is doing to AI development right now.
The curve that hasn’t bent
METR tracks the length of tasks AI can reliably complete on its own. That horizon is doubling roughly every four months — up from every seven. Anyone can check this in public data.
Task horizon — how long a job AI can handle solo
Each model handles dramatically longer tasks than the one a year before. The line keeps going up.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL500_.jpg "When AI Builds Itself: Inside Anthropic’s Evidence on Recursive Self-Improvement 3")
Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Two kinds of work, one persistent gap
Building a frontier model splits into engineering and research. Across both, the pattern is the same — and so is the one thing AI still can’t do well.
Code, infrastructure, training
Claude can take an underspecified problem and find a method. Humans supply the goal; they no longer need to supply the method.
Which experiments, what they mean
Claude can match or outperform skilled humans at executing a well-specified experiment. But choosing which experiment still needs a human.
The same ladder Anthropic employees climb with experience
AI VoiceWriter – Smart Dictation & AI Writing Assistant for Windows & Mac | USB Dongle & Mobile App for Voice Input, Proofreading, Rewriting & Multilingual Support
🎙️ Hands-Free Voice Typing for Windows & Mac – Powered by iOS & Android dictation technology, AI VoiceWriter…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Watch the human share shrink, rung by rung
Walk up the four stages of AI development. At each, the human/AI split shifts — and the real internal numbers show exactly where AI has reached parity, gone superhuman, or still trails. Tap a rung.
The human role across the development loop
The doing now costs almost nothing in human time. What’s left is the deciding.
reBot B601-DM Assembled Robotic Arm Kit with Gripper, 6+1 DoF Open-Source Robot Arm, Python SDK and ROS1/ROS2 Compatible for AI Robotics, STEM Education, Research and Development
Fully Assembled 6+1 DoF Robotic Arm: reBot Arm B601-DM comes as a pre-assembled robotic arm kit with gripper,…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Agents ran an open research project end to end
April 2026: the first demonstration of Claude running an open-ended research project from hypotheses to findings — on a real AI-safety problem.
Can a weaker model reliably supervise a stronger one?
Agents were left to solve it: proposing hypotheses, testing them, sharing findings across parallel agents, iterating. Measured against the gap between a “floor” (weak supervisor alone) and “ceiling” (strong model trained on correct answers).
(humans: ~23% in a week)
· ~$18,000 compute
the agents themselves
Pushing and Merging Code
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Picking a better next step than the human
Real research sessions where a human took a wrong turn. Models saw only the work before the detour and proposed a next step; a judge that knew the outcome scored them. The day-to-day of research is this chain of next-step calls.
“Can the model pick a better next step than the human?”
Share of moments where the model’s next step was judged better. The amber line is the practical ceiling (an ideal answer that could see the whole session).
It depends on whether the trend continues — and what we do
The piece refuses a single prediction. It lays out three scenarios, and is clear about which it finds most likely.
The exponentials turn out to be S-curves
Maybe taste can’t be scaled into existence; maybe the constraint is the supply chain — chips, grid, interconnect — not intelligence. Even so, the world still changes: Glasswing’s Mythos found 10,000+ critical vulnerabilities in weeks, and a 100-person firm does the work of 1,000.
included for completeness · they doubt itDevelopment automates; humans still steer
100-person companies doing the work of tens of thousands — revolutionary, but turnable to harm (population-scale surveillance, tailored manipulation). Bound by Amdahl’s law: speeding one part shifts the bottleneck — which is exactly why human code review became Anthropic’s new chokepoint.
★ they think we’re likely heading hereAI designs and refines its own successors
Progress paced only by compute. Humans move to oversight of an expanding “virtual lab.” The future they understand least — especially whether alignment holds, or whether rare misalignments compound as models build successors, until control slips.
the one they’re most uncertain aboutBuild the option to slow down — verifiably
The piece ends on policy, not product. A unilateral pause just changes who leads; what’s missing is the ability to verify others have actually slowed.
Why a credible pause is hard — and worth building toward
A slowdown that only lets the least cautious catch up leaves everyone less safe. So the goal is the option: systems that let frontier labs verify others have genuinely stopped. Anthropic says if such systems existed and peers paused verifiably, it expects it would too.
Detection beats verification — and even that’s tough
Training runs are easier to conceal than missile silos, inputs are general-purpose, and whoever continues while others pause inherits the lead.
We’ve done it before — slowly
Regimes like the INF Treaty built verification and trust over decades. The authors’ blunt line: “We don’t have that long.”
Reading it in proportion
- This is one lab’s account of its own internal data — much previously unreported, not independently audited.
- The soft spots are stated in the original: lines-of-code overstates productivity; the self-reported 4× is probably high; the headline research result didn’t transfer to production scale; the next-step test used cherry-picked moments.
- “More autonomous” is not “fully autonomous” — every standout result still had a human framing the problem and defining success.
- That the authors surface these caveats themselves — against their own incentive — is part of what makes the document serious.
Implications of AI-Driven AI Research and Development
This evidence suggests that AI systems are already transforming the pace of AI development, which could lead to a rapid cycle of self-improvement if the decision-making bottleneck is removed. Such a shift could accelerate technological progress beyond current expectations, raising important questions about control, safety, and regulation in AI development. The findings challenge the assumption that recursive self-improvement is purely theoretical, indicating it may be closer than previously thought.
Recent Trends in AI Self-Development Capabilities
Over the past few years, AI benchmarks have shown consistent improvements in models’ ability to perform increasingly complex tasks, with the pace of progress accelerating. Public data from benchmarks like METR and SWE-bench reveal a pattern of exponential growth in AI capabilities, supporting the idea that AI is becoming more autonomous in its research activities.
Anthropic’s internal data, shared in their recent report, provides a rare glimpse into the actual internal progress, showing AI’s rising role in coding, experimentation, and problem-solving within labs. This internal evidence complements public benchmarks, which are limited to measuring task performance but do not capture the internal pace of development.
“The data indicates that AI is already automating significant parts of the research process, and if decision-making bottlenecks are overcome, we could see a rapid loop of self-improvement.”
— Thorsten Meyer, author of the report
Unresolved Questions About Autonomous AI Self-Improvement
It remains unclear whether AI can fully automate the decision-making processes involved in research, such as choosing the most promising problems or evaluating results without human input. The report emphasizes that the bottleneck of human judgment is still significant, and whether it can be automated at scale is an open question. Furthermore, the timeline for achieving full recursive self-improvement is uncertain, with experts warning that technical, safety, and ethical challenges could slow progress or prevent it altogether.
Next Steps in Monitoring AI Autonomous Development
Researchers and industry stakeholders will closely watch internal progress at labs like Anthropic, as well as benchmark trends, to assess whether AI systems are approaching the critical threshold for autonomous self-improvement. Regulatory and safety discussions are likely to intensify as the pace of capability growth accelerates, emphasizing the need for frameworks that can adapt to rapid technological shifts. Future research may focus on automating higher-level decision-making in AI development, testing whether the bottleneck can indeed be broken.
Key Questions
What is recursive self-improvement in AI?
Recursive self-improvement refers to an AI system’s ability to improve its own capabilities autonomously, potentially leading to rapid, exponential growth in intelligence and performance without human intervention.
How does Anthropic measure AI’s progress in automating research tasks?
Anthropic uses public benchmarks like METR, SWE-bench, and CORE-Bench, alongside internal data, to track AI’s ability to perform tasks such as coding, experiment design, and problem-solving, observing rapid improvements over recent years.
Is full autonomous self-improvement already happening?
No, the report indicates that while AI is automating many research tasks, the critical decision-making step—choosing goals and evaluating results—remains human-controlled. Full autonomous self-improvement is still a future possibility.
Why does this development matter for AI safety?
If AI systems begin to autonomously improve themselves at a rapid pace, it raises questions about control, alignment, and safety, emphasizing the need for careful monitoring and regulation as capabilities advance.
What are the next milestones to watch for?
Key milestones include AI systems autonomously designing their own research agendas, automating goal selection, and achieving breakthroughs in internal decision-making processes that currently require human input.
Source: ThorstenMeyerAI.com
