Posted on 06/05/2026 8:20:55 AM PDT by ProtectOurFreedom
For most of AI’s history, humans drove every step in its development cycle. But at Anthropic, we are delegating a growing share of AI development to AI systems themselves, which is speeding up our work.
Taken far enough, and given enough compute, that trend points to an AI system capable of fully autonomously designing and developing its own successor. This is called recursive self-improvement. We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for.
Using public benchmarks and previously unreported data from within Anthropic, The Anthropic Institute is showing that AI is already accelerating the development of AI systems. To take just one example: today, Anthropic engineers on average ship 8x as much code per quarter as they did from 2021-2025.
The rate at which AI models improve is accelerating. The length of tasks that they can reliably complete on their own has been doubling roughly every four months, up from an earlier trend of doubling every seven months. In March 2024, Claude Opus 3 could complete software tasks that take humans about four minutes to complete. A year later, Claude Sonnet 3.7 managed tasks that took about an hour and a half. A year after that, Claude Opus 4.6 managed 12-hour tasks.1 If this trend holds, tasks that take a skilled person days could come into range this year. In 2027, AI systems could be capable of tasks that take a person weeks.
The same pattern appears on coding and research benchmarks. Benchmarks measure the performance of models in a given domain, and they’re “saturated” when models achieve close to 100% performance.2 SWE-bench is a standard test of real-world software engineering: it hands a model an actual open-source codebase and a real bug report, and asks it to write a code change that fixes the issue and passes the project’s own tests. Models have gone from scoring in the low single digits to saturating the benchmark in two years.
CORE-Bench tests whether a model can reproduce existing research, a prerequisite for them to conduct original research. It gives an AI model the code and data behind a published paper, and asks it to rerun everything and confirm it can replicate the paper’s results. AI systems went from succeeding at reproducing the results roughly 20% of the time in 2024 to saturating the benchmark fifteen months later. METR, which runs the benchmark measuring how well models can complete long-duration tasks, found that Claude Mythos Preview could work for “at least” 16 hours and was “at the upper end of what [METR] can measure without new tasks.”
Public benchmarks say a lot about the capabilities of these systems. But they can’t reveal the impact AI systems are having on speeding up AI development itself. For that, we need direct evidence from within AI companies like Anthropic.
Across both engineering and research, the picture is consistent. In engineering, Claude can be handed an underspecified problem and figure out how to solve it; humans supply the goal, but they no longer need to supply the method. In research, Claude can already match or outperform skilled humans at executing a well-specified experiment.
As of May 2026, more than 80% of the code we merge into Anthropic’s codebase was authored by Claude.3 Before Claude Code launched in research preview in February 2025, this number was in the low single digits. That shift also shows up in the amount of output per engineer. Lines of code merged per engineer per day stayed constant through Anthropic’s first four years (2021-2024), then began to climb upward in 2025 when Claude began to run code rather than just suggesting it for an engineer to copy and paste. The slope steepened again in 2026 when models began to work autonomously over longer time horizons. These two inflection points are shown in the chart below. In the second quarter of 2026, the typical engineer was merging 8× as much code per day as they were in 2024.4 This is because much of the code is written by Claude, with the engineer directing and reviewing, rather than typing it themselves.
The code that Claude writes is “good” and improving. “Good code” means two things: it works, and it is written in a manner that allows another engineer to understand it and build upon it. On the first criterion, the evidence is clear. The rate at which Anthropic staff correct, redirect, or take over mid-task from Claude has been falling steadily for a year, including on the most complex and open-ended tasks. This means problems with no clear specification, where the engineer isn’t sure what the answer looks like. This is evident in Claude’s success rate over time on tasks of different difficulties, as shown in the graph below. Claude writes code that works.
On the most open-ended tasks, Claude’s success rate reached 76% in May 2026, up 50 percentage points in six months. To give an example of tasks in this difficulty tier, a routine upgrade began crashing tens of thousands of training jobs. An engineer pointed Claude at the live incident with little more than some text content and cluster access. Working through the running jobs and testing one environment setting at a time, Claude isolated the single obscure debugging flag that was triggering the crash, reproduced it reliably, and confirmed a fix. In about two hours, Claude delivered what would normally be two to three days of work.
The second criterion is writing code that another engineer can understand and build on. Here the gap between humans and AI persists, but is closing fast. There isn’t full consensus among staff at Anthropic, but many believe that the Claude-written code was still worse in quality than human-written code at Anthropic in late 2025, and is roughly at parity today. We expect it to be better within the year.
This has changed the way that Anthropic now reviews its own code. Proposed changes to our codebase are now read by an automated Claude reviewer that looks for bugs, security flaws, and other defects before it can merge. Using this tool, we ran a retrospective analysis, and found that an automated Claude review of every change to our codebase would have caught roughly a third of the bugs behind past incidents on claude.ai before they ever reached production. The engineers who wrote that code are among the best in the world at building these systems. Claude is now catching the mistakes that they missed.
An area of human comparative advantage, for now, is research taste and judgment, including choosing which problems matter, which results to trust, and when an approach is a dead end.
Even if we suppose that Claude never achieves good research taste, a conservative reading of our evidence still implies compounding acceleration. If humans spend most of their time on the single-digit fraction of work that is direction-setting, while Claude handles the rest, that means each engineer or researcher is steering far more work than before. The evidence we see suggests that people at Anthropic are both moving faster and covering a broader surface. In practice, this means that AI already makes Anthropic move much faster than it did before the advent of effective AI tools.
The less conservative reading is that the early evidence on Claude’s improving research judgment—narrow as it is today—is an indicator that this capability is improving as well. “Research taste” might be just another AI capability that AI systems fail at for a time, then get good at. We’ve seen a similar pattern with other qualitative skills, like AI systems being able to explain why a joke is funny, demonstrate theory of mind, and solve linguistic riddles.
1. The trend stalls, but today’s AI capabilities are widely diffused.
2. AI labs continue to see compounding efficiency gains.
3. AI systems themselves become capable of full recursive self-improvement, and begin building their successors.
We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology. The Anthropic Institute will conduct research—in collaboration with many others—and take actions to help build the systems that a credible slowdown or pause would require. These systems would enable frontier AI developers to verify that others globally have actually stopped or slowed, and that a bad actor could not use the auspices of a coordinated slowdown to jump ahead in secret. If such systems existed, we expect that we would slow down or temporarily pause, if other developers at or near the frontier also did so in a verifiable manner.
A meaningful slowdown or pause would require multiple well-resourced labs at or near the frontier, in multiple countries, agreeing to stop under the same conditions. It would also require that each can verify that the others have actually stopped. Due to the unique characteristics of AI systems, the detectability (a lower standard than verifiability) element of this arms control problem is much more challenging than with other technologies. Training runs are far easier to conceal than missile silos, their inputs are general-purpose, and the incentive to defect quietly is enormous, because whoever continues while others pause could inherit the lead. A credible pause also has to specify what triggers it, what lifts it, and who adjudicates.
None of this is necessarily impossible in principle—the world has built verification regimes for other complex technologies (e.g., the Intermediate-Range Nuclear Forces Treaty)—but those regimes took decades to build both the infrastructure and the trust. We don’t have that long. A unilateral pause by one lab, by contrast, is achievable immediately, but accomplishes much less: it would change who the front-runner is, but it would not create the wider deliberative process that is currently missing.
In the coming months, we will organize conversations where policymakers, researchers, civil society, and other AI companies can help answer some of the questions this piece raises, especially around full recursive self-improvement and how to create better options for coordination and deliberation. We’ll publish what comes out of it. The window to investigate the questions together is here, and people outside AI companies should be involved in this deliberation.
Dear FRiends,
We need your continuing support to keep FR funded. Your donations are our sole source of funding. No sugar daddies, no advertisers, no paid memberships, no commercial sales, no gimmicks, no tax subsidies. No spam, no pop-ups, no ad trackers.
If you enjoy using FR and agree it's a worthwhile endeavor, please consider making a contribution today:
Click here: to donate by Credit Card
Or here: to donate by PayPal
Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794
Thank you very much and God bless you,
Jim
Thank you for posting. I just shared this with some of my colleagues.
My pleasure. Glad you and [hopefully] your colleagues find it useful. I find it to be a very well reasoned, factually supported paper and thought-provoking from those people in the trenches who know this better than anybody. Even they have wide-ranging hypotheses about the future ranging from “we’ll hit the wall soon” to “all bets are off.” The best minds just don’t know.
If you can, bring us feedback from your colleagues.
Can AI create/build/operate a plant to manufacture transformer coils or is it just a Q&A, code generation, and design tool?
“Can AI create/build/operate a plant to manufacture transformer coils”
I used to be very skeptical it could ever do much in the physical world, but, with the enormous advances in robotics, I have to admit that I’m rethinking that.
Is machine replacement for humans five years away? Ten? Twenty five? Who knows what the world will look like in 100 years. I firmly believe AI + robotics is going to make the next 100 years the most rapid transformative period in all human history. And we’ve seen staggering changes in the most recent two 100 year periods.
When that happens the social consequences will be enormous!
Particularly so for our ill-educated, indifferently educated by K12 workforce!
“our ill-educated, indifferently educated by K12 workforce!”
And unemployed with no meaning in life. Tens of millions of bored criminals. “Mad Max” here we come.
Agree!
We think criminality and drug\alcohol use are bad now, just wait!
Reily wrote, “When that happens, the social consequences will be enormous! Particularly so for our ill-educated, indifferently educated by K12 workforce!”
When robotics and recursive AI reach this level, human labor will have no value. The successful class of people will be those who can do what only humans can do: innovate from new ideas. Jobs that determine “truth” and “beauty” will be all that is left. These require educated, aware, and skilled innovators.
This will result in a completely new class division of people: Those who can create and those who can’t. Labor capital value will approach zero, while the financial capital value that can be invested in the surviving companies will go parabolic. At some point, this division will result in a revolution.
Don’t disagree with the current trajectory.
Can we educate “educated, aware, and skilled innovators” with unionized government schools or even private schools?
I just closed my machine shop last month after 20 years. I can’t compete with what the Chinese can do. And updating my machinery would have cost a million bucks I don’t want to spend.
Having played around with generative AI design software (Autodesk) it will create awesome designs that are nearly impossible to manufacture in the real world.
Now, Im slowly playing with AI to handle mundane tasks, travel ideas, book keeping, and writing code for automation equipment. Im not sure how I can leverage it to make money for me but it sure makes those tasks quick.
We will soon have perfect paperclips!
Yeah, we’re screwed.
No, I’m joking. Based on what I’ve seen, AI will spin itself into ever-diminishing fractal-like spirals of uselessness.
It takes a human (so far) to properly direct AI, and it seems to be a very iterative process.
Having worked in a factory as a transformer coil winder and later a software developer/engineer, I look at AI as an extension of what we called (in my time) knowledge-based and expert systems based on a set of rules, but I can’t see robots as so smart that they can design/build/operate a factory, and automatically distribute a product (which is probably the easiest part).
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.