Step 6: tool results

provenance_read_source

{
  "max_chars": 7000,
  "source_id": "ece6f61a-093f-4e2c-ba9d-801d19e6370e"
}

{"success":true,"source":{"id":"ece6f61a-093f-4e2c-ba9d-801d19e6370e","url":"https://forecastingresearch.substack.com/p/how-close-is-recursive-self-improvement","title":"How Close Is Recursive Self-Improvement?","source_type":"blog_post"},"origin":"stored","total_chars":11319,"offset":0,"content":"Subscribe Sign in\n\nHow Close Is Recursive Self-Improvement?\n\nOur new report reviews the evidence, turns key disagreements into forecasting questions, and gives initial LLM forecasts.\n\nForecasting Research Institute\n\nNov 10, 2026\n\n23\n\n4\n\n3\n\nShare\n\nPost written by Simas Kučinskas and Josh Rosenberg\n\nAI companies increasingly say that they may soon achieve recursive self-improvement (RSI). With RSI, current AI systems help develop future AI systems in a recursive loop. RSI could dramatically accelerate AI capability progress. Importantly, this acceleration could happen inside AI companies before outsiders can notice.\n\nAI companies now say that this loop is starting. In an update to its IPO filing, Anthropic claimed that, based on expected future models, it now meets its definition of “dramatically accelerated progress attributable to AI.” OpenAI said that it reached its goal of an “automated AI research intern” in September 2026, and it targets an automated AI researcher by March 2028.\n\nOf course, AI companies are not necessarily neutral observers. So how close is RSI really? In our new report , we provide a comprehensive overview of AI companies’ reported data, public benchmarks, and existing forecasts . 1 We also turn the key disagreements into forecasting questions and provide initial forecasts from an ensemble of frontier LLMs .\n\nThis post summarizes what we found.\n\nAI is a highly capable research assistant, but humans still supply most of the research judgment\n\nThe data reported by AI companies show that AI does a large and fast-growing share of AI R&D.\n\nAt Anthropic, Claude wrote over 90% of contributions to the codebase in September 2026. At OpenAI, AI agents did 3.1 workdays of work for each human workday in August 2026, up from about 0.5 in May 2026.\n\nHowever, humans still supply most of the research judgment, such as choosing which research areas to focus on. For instance, no measured part of Anthropic’s AI R&D is fully autonomous . Anthropic also states that “large performance gaps persist when it comes to Claude exercising judgement in choosing goals in both engineering and research.” At OpenAI , AI completed 57% of 2–4-hour tasks with no human intervention in July 2026 (up from 28% in January) and 35% of 8–16-hour tasks (up from 10%).\n\nPublic evidence points the same way\n\nAI agents do well on well-defined machine learning (ML) benchmarks.\n\nOn PostTrainBench , where agents turn small base models into useful assistants, the best agent scores 45.6, against 51.1 for the official instruct models. On WeirdML , where agents build ML models for small, unusual problems, GPT-6 Astra scored 42.2 (out of 100) in September 2026, up from 15.0 for GPT-5.6 Sol in July 2026. On TasteVal , where a model chooses and interprets AI R&D experiments, Claude Opus 5.5 reaches the score of human experts with 2.3× less compute.\n\nHowever, AI agents struggle with open-ended research. On InnovationEval , where agents try to reinvent a published ML innovation they have not seen, GPT-5.6 Sol recovered 15% of the innovation’s gain. (GPT-6 Astra scored 63% but had likely memorized the original paper.) In the CRUX 2 open-world evaluation, AI agents got the research question of an unpublished AI paper and had to write their own paper on it. The original paper’s authors rejected all four AI-written papers, including two from a follow-up study with newer models, more time, and a larger budget.\n\nThere is some internal evidence of a potential acceleration\n\nAI capability progress has not clearly accelerated on public measures. Since September 2024, the frontier of the Epoch Capabilities Index (ECI) has risen by about 14 points a year . Here, the ECI frontier is the highest ECI score of any publicly available model by a given date.\n\nThat said, there is some internal evidence from frontier AI companies of a potential acceleration. In the Claude Opus 5.5 system card , a preliminary METR assessment estimated “~1.5X overall acceleration in capabilities due to AI” inside Anthropic, with “perhaps 30% chance of 2X.” At OpenAI , experiments per active experimenter hit an all-time high in August 2026, about 1.6× the 2025 level. However, OpenAI notes that “our available compute has also grown significantly since 2025.”\n\nIn a report published after our November 3 data cutoff, METR now estimates that capability progress at Anthropic is about 2× its earlier rate, and that around 80% of this acceleration is due to AI. This estimate applies to Anthropic only and includes unreleased models. In addition, part of the acceleration may reflect Anthropic catching up to the frontier, and Anthropic redacted much of the evidence that METR found most informative.\n\nBroad acceleration needs three contested steps\n\nExisting forecasts about the likely trajectory of RSI diverge widely.\n\nOn a simple extrapolation of current trends, mostly using company-reported metrics, AI could take over most AI R&D tasks by 2027–2028, in line with OpenAI’s internal goal of an automated AI researcher by March 2028. A key uncertainty is whether these trends continue, as the remaining non-automated tasks increasingly require research judgment. Forecasts of full automation of AI R&D range from 2027 (insiders at AI companies and the AI Futures Project’s AI 2027 scenario) to 2052 (the AI Impacts survey of ML researchers).\n\nIn our report, we cover the key points of debate, including the views of researchers who expect RSI very soon and those who are more skeptical. For example, the AI Futures Project’s AI 2027 scenario depicts a superhuman AI researcher in 2027. In contrast, Arvind Narayanan and Sayash Kapoor expect gradual automation of AI development instead of “a singular, discontinuous moment when recursive self-improvement is achieved.”\n\nWhy do people disagree? Broad acceleration through automated AI R&D needs three steps to happen in turn , as shown in the diagram below:\n\nAutomation: Is AI R&D automated?\n\nAcceleration: Does AI capability progress accelerate sharply?\n\nBreadth: Are capability gains broad?\n\nEach step is contested. The main cruxes are whether AI develops research taste, whether diminishing returns or bottlenecks such as compute, data, or regulation cap the acceleration, and whether gains spread beyond AI R&D and easily verified tasks.\n\nFrontier LLMs forecast about a 40% chance of a “fast year” by 2030\n\nWe turned these cruxes into 18 forecasting questions.\n\nTo measure acceleration, we propose a forecasting question on the likelihood of a “fast year”: any 12 months in which the ECI frontier rises by 30 points or more. This increase is about 2× ECI’s pace since September 2024. The largest rise in any 12-month window (as of our data cutoff of November 3, 2026) is 21 points.\n\nTo get an initial set of forecasts, we asked an ensemble of four frontier LLMs: Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.1, and GPT-6 Astra. Each LLM did its own web research and answered every question five times. We used the forecasting pipeline of our Au","truncated":true,"next_offset":7000,"facts":{"authors":["Forecasting Research Institute"],"publisher":null,"published_date":null,"doi":null,"archived_url":null,"events":[],"versions":[],"copy_of":[]},"note":"More text follows; call again with offset 7000 to continue."}