{"id":2711,"date":"2026-11-10T16:44:22","date_gmt":"2026-11-10T16:44:22","guid":{"rendered":"https:\/\/forecastingresearch.org\/?post_type=research&p=2711"},"modified":"2026-11-10T18:03:44","modified_gmt":"2026-11-10T18:03:44","slug":"recursive-self-improvement-v1","status":"publish","type":"research","link":"https:\/\/forecastingresearch.org\/research\/recursive-self-improvement-v1","title":{"rendered":"How Close is Recursive Self-Improvement? A Review of the Evidence"},"content":{"rendered":"\n

Summary<\/h2>\n\n\n\n

DATA CUTOFF: NOV 3, 20261<\/a><\/sup><\/p>\n\n\n\n

AI companies increasingly state that their models may soon be capable of recursive self-improvement<\/em> (RSI), which could dramatically accelerate AI capability progress. How close is RSI? We review the evidence, existing forecasts, and key disagreements, turn the disagreements into forecasting questions, and give initial forecasts from an ensemble of frontier large language models (LLMs). As of early Nov 2026, AI acts as a highly capable research assistant, but humans still supply most of the research judgment. AI capability progress has not clearly accelerated on public measures, although there is some internal evidence from AI companies indicating a potential acceleration. On a simple extrapolation of current trends, mostly using company-reported metrics, AI could take over most AI R&D tasks by 2027\u20132028, in line with OpenAI\u2019s target of an automated AI researcher by Mar 2028. A key uncertainty is whether these trends continue as the remaining non-automated tasks increasingly require research judgment. To measure acceleration, we propose a forecasting question on the likelihood of a fast year<\/em>: any 12 months in which the frontier of the Epoch Capabilities Index rises by 30 points or more, about twice its pace since Sep 2024. The LLM ensemble forecasts about a 40% chance of a fast year by 2030. If a fast year happens, the ensemble gives about a 70% chance that AI automation of AI R&D caused it.<\/p>\n\n\n\n

1. Introduction<\/h2>\n\n\n\n

How quickly will AI improve between now and 2030?<\/p>\n\n\n\n

There are three broad possibilities: AI progress could slow down, continue at its current pace, or accelerate. A particularly important scenario is acceleration because AI systems help build future AI systems<\/strong>. The International AI Safety Report 2026 (IASR) notes that \u201cpositive feedback loops \u2013 such as AI systems contributing to AI research \u2013 could accelerate progress<\/em>\u201d and \u201cdecades of progress could happen in years<\/em>\u201d (IASR<\/a>, \u00a71.3). This positive feedback loop is often called recursive self-improvement (RSI)<\/strong> (Yudkowsky, 2008<\/a>): AI systems automate AI R&D and build better successor systems in a recursive loop (see Figure 1).<\/p>\n\n\n\n

\"fig-01\"
Figure 1:<\/strong> The RSI loop, in which each generation of AI systems helps develop the next.<\/figcaption><\/figure>\n\n\n\n

RSI could bring about large benefits as well as substantial risks. A Sep 2026 report by AI researchers including Geoffrey Hinton, Yoshua Bengio, OpenAI chief scientist Jakub Pachocki, and Anthropic cofounder Jack Clark argues that RSI could bring AI\u2019s benefits forward by years. However, RSI could also pose extreme risks, as humanity may fail to keep up with the growth in capabilities or lose oversight of advanced AI systems (Chan et al., 2026<\/a>).<\/p>\n\n\n\n

Recent reports review RSI with a focus on its risks and on policy responses (Chan et al., 2026<\/a>; Sarma et al., 2026<\/a>). This review is complementary and forward-looking. We summarize the current evidence and the key disagreements, turn these disagreements into forecasting questions, and give an initial set of forecasts from an ensemble of frontier LLMs.<\/strong> Simultaneously, we are asking a group of top human forecasters and AI experts to make forecasts on RSI, using this document as background material. We plan to add their forecasts in a later update of this report.<\/p>\n\n\n\n

How to use this document:<\/strong><\/p>\n\n\n\n