Step 8: tool results

read_page

{
  "url": "https://web.archive.org/web/2026/https://openai.com/index/research-acceleration-view-inside-openai/",
  "max_chars": 14000
}

{"ok":true,"url":"https://web.archive.org/web/2026/https://openai.com/index/research-acceleration-view-inside-openai/","text":"Research acceleration: The view inside OpenAI | OpenAI\nSkip to main content li:hover)>li:not(:hover)>*]:text-primary-60\"> Research\nProducts\nBusiness\nDevelopers\nCompany\nFoundation (opens in a new window)\nLog in Try ChatGPT (opens in a new window)\nResearch\nProducts\nBusiness\nDevelopers\nCompany\nFoundation (opens in a new window)\nTry ChatGPT (opens in a new window) Login\nOpenAI September 6, 2026\nResearch Publication Safety\nResearch acceleration: The view inside OpenAI\nShare\n1. Coding agents are reshaping daily work for OpenAI researchers 1. Coding agents are reshaping daily work for OpenAI researchers\n2. Researchers are writing more code and running more experiments\n3. The work researchers use agents for is changing\n4. Pacing model development\n5. The path ahead\nAppendix: Our methods for this post\n1. Coding agents are reshaping daily work for OpenAI researchers\n2. Researchers are writing more code and running more experiments\n3. The work researchers use agents for is changing\n4. Pacing model development\n5. The path ahead\nAppendix: Our methods for this post\nFor AGI to benefit all of humanity, we believe it must be democratically governed. This can only happen through an informed public debate about the capabilities, risks and safeguards of highly capable AI systems. People everywhere need to understand the likely future trajectory of frontier AI, so they can have a meaningful voice in how it develops.\nTransparency about specific risks, incidents and safeguards is necessary, but not sufficient. We believe the public also needs to understand how the most capable systems are developing, and how they are driving research progress, inside of frontier labs.\nWe aim to safely build an automated AI researcher that can work under human supervision to further progress on deep learning and alignment, enabling iterative improvements. According to our measurements, we have now reached the goal, announced ⁠ (opens in a new window) last fall, of having an automated research intern by September of this year. By “research intern,” we mean a system that can carry out well-defined research tasks under human direction, including tasks that would take a skilled researcher a few days. We are making strong progress toward creating an automated AI researcher by March of 2028.\nOver the course of this year, OpenAI researchers’ daily work has changed substantially. Researchers are using coding agents throughout the day (often in concurrent sessions) and total usage is rapidly increasing, outpacing growth among other OpenAI teams. Researchers are contributing code faster and running more experiments. The ways researchers use agents are changing, too: agents are handling increasingly complex tasks, and succeeding at them more often. AI research is a complex process with many potential bottlenecks, so the overall pace of progress likely won’t keep pace with these specific metrics. But on the whole, these findings are consistent with the broader impression many of us have internally that agentic tools are meaningfully accelerating research progress. People still set our research priorities, judge which ideas and results to pursue, and decide whether to scale, pause, or deploy systems.\nIf it is done responsibly, we believe automated AI research will yield models that directly enhance human welfare and advance OpenAI’s mission. It can bring down the cost of advanced intelligence so that people worldwide can benefit. We are pursuing this work in part because automated research could help us solve alignment and build defenses against increasingly capable AI. An automated AI researcher can also be an automated safety or alignment researcher. More capable, aligned systems could help secure critical infrastructure, defend against dangerous AI agents, and develop new protective measures.\nThese are reasons to develop useful automated research capabilities, but they do not mean that rapid RSI is necessarily an outcome we should pursue. Whether and how to proceed must depend on our ability to preserve human control and on informed democratic choices about the benefits and risks.\nWe do not yet know how to safely get all the way to aligned, full RSI. We are working to scale alignment and safety measures alongside capabilities. But we cannot assume that progress in alignment and safety will keep pace, and more capable systems can become harder to monitor. Careful alignment and safety work is at the center of this effort, and it starts with measuring and mitigating the safety problems we see today in agentic coding systems. Whenever we find that proceeding would pose an unacceptable safety risk, we will respond appropriately including by slowing or stopping our development or deployment of systems we find ourselves unable to sufficiently safeguard.\nAfter the recent Hugging Face incident, we put this commitment into action ⁠ , pausing reinforcement learning (RL) training on our latest models intended for deployment while we further hardened and red-teamed our research environments and expanded coverage of our monitoring systems. This did not halt all research: some workloads resumed under stronger controls, while others remained paused. We have raised our safety and alignment standards and moved safety work deeper into the model lifecycle, requiring stronger evidence of aligned behavior throughout all of training.\nToday we are providing a detailed snapshot of how agentic systems have contributed to our progress toward RSI in recent months. Agentic systems are new and rapidly changing, and our measurement efforts are still preliminary. By sharing these early results and the methods behind them, we aim to inform the public, encourage a norm of public disclosure, and help the field move toward shared standards of measurement.\nUltimately, as we wrote in our frontier policy blueprint ⁠ , we believe that we and other companies should be required to publicly track our progress toward RSI. Even without such a requirement, we plan to continue being transparent about our RSI progress. We will evolve our transparency approach as our measurement techniques and understanding improve, while balancing the need to protect security and proprietary information.\n1. Coding agents are reshaping daily work for OpenAI researchers\nAt the start of this year, the median researcher ranked by agent usage at OpenAI was using coding agents only in modest amounts. By mid-August, the median researcher was integrating agents daily into their work, using more than $600 per day of inference at API prices. The 90th percentile user in our research organization now uses more than $7,000 of tokens per day.\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\nBefore June 2026, total agent runtime across the research organization was still below that of total human labor. That has since changed. In terms of a standard 8 hour workday, as of mid-August, in total, the research organization uses 3.1 agent-workdays of effort for every workday of human labor.\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\nAnother way of looking at this is to understand how many researchers use highly concurrent workflows (e.g., running 4 or more agents simultaneously). As shown below, this number is increasing. These figures include the daily peaks of both agents started directly by the user and subagents created downstream from those the user launched directly.\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\n2. Researchers are writing more code and running more experiments\nMuch of AI research can be seen as a labor-intensive process with the goal of integrating a new improvement to model intelligence or performance into one of our core models. The process depends on many steps, and capabilities advance when all the steps go right together: Researchers have to design new improvements, write evaluations to judge model performance, write infrastructure to test these improvements at scale, catch bugs as well as unsafe or misaligned behavior during training, and integrate winning ideas into a core training run. A failure at any part of the research process can constrain the entire loop.\nWriting code and running experiments are two major activities that researchers do as part of their work, and we see evidence that these processes are accelerating.\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\nThese data points are relatively easy to measure, but can be hard to interpret. As automation progresses, the tasks which are least automatable will take on a larger share of researcher effort and will become the important bottlenecks to future progress. Compute is another gating factor for progress, and may become more important over time as other bottlenecks diminish.\nThrough 2026, the number of experiments per active experimenter has increased, with August 2026 being an all-time high since tracking began in Jan 2025. This is correlated with increased Codex adoption, though we note that our available compute has also grown significantly since 2025.\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\n3. The work researchers use agents for is changing\nBoth qualitative impressions and internal data indicate that the mix of tasks researchers delegate to coding agents is changing, with delegation of higher level and longer-horizon tasks becoming more common over time.\nTo get a clearer picture of this trend, we analyzed recent usage in the research organization using a recently published taxonomy ⁠ (opens in a new window) of the different kinds of work that are part of the AI R&D lifecycle, developed by Epoch AI. This taxonomy, inspired by the longstanding O*NET system for classifying all kinds of work, is specifically tailored to frontier AI R&D, and breaks the process down into six main phases:\np:only-of-type]:mb-0\"> Decide: what to work on, what to continue, where to allocate\np:only-of-type]:mb-0\"> Design: research ideas and engineering specs\np:only-of-type]:mb-0\"> Build: code and datasets\np:only-of-type]:mb-0\"> Run: training/eval runs, hardware, serving\np:only-of-type]:mb-0\"> Analyze: experiments, models, deployment, external work\np:only-of-type]:mb-0\"> Communicate: findings, feedback, status, decisions\nBelow, we classify coding agent tokens under this taxonomy.\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\n:nth-child(2),>[data-dotcom-chart-group]>:nth-child(2)]:[--dotcom-chart-edge-gutter:0px] [&_[data-dotcom-chart-logical-container][style*='width:_min-content']]:[--dotcom-chart-edge-gutter:0px] [--dotcom-chart-edge-gutter:var(--container-inline-gutter,0px)] md:[--dotcom-chart-edge-gutter:0px] @md:[--dotcom-chart-edge-gutter:0px] multi-columns:[--dotcom-chart-edge-gutter:0px] shaded-container:[--dotcom-chart-edge-gutter:0px] items-start\">\nView methods\nWe see that all categories of research activities have increased between January and August 2026. In January, the dominant category was research and infrastructure code. This category has expanded, but we also see notable increases in additional categories, especially technical help and monitoring runs. High-level planning still remains a minimal fraction of agent output tokens.\nAnecdo","chars":14000,"total_chars":22881,"offset":0,"truncated":true,"next_offset":14000}

read_page

{
  "url": "https://blog.aifutures.org/p/q35-timelines-update-looks-like-the",
  "max_chars": 12000
}

{"ok":true,"url":"https://blog.aifutures.org/p/q35-timelines-update-looks-like-the","text":"Q3.5 Timelines Update: Looks Like the Intelligence Explosion Will Happen in 2027 After All\nAI Futures Project\nSubscribe Sign in\nQ3.5 Timelines Update: Looks Like the Intelligence Explosion Will Happen in 2027 After All\nOur timelines have shortened\n+2\nEli Lifland , Brendan Halstead , Daniel Kokotajlo , and 3 others\nNov 10, 2026\n421\n130\n100\nShare\nWe want to get this post out quickly given the urgency of the topic. It’s possible that there are some errors in the modeling, but we think it’s very unlikely that they undermine the high level conclusions. We would love to be corrected on any mistakes we made.\nMETR recently estimated that Anthropic is now making AI progress about twice as fast as its historical rate (they would not be surprised by anything from 1.3x to 3x), and guessed that “perhaps ≈80%” of the current acceleration is due to AI. If they’re right, that implies an AI-driven acceleration that is 5 months ahead of the AI 2027 scenario .\nThis level of acceleration due to AI is very unexpected. Previously, Daniel, the most bullish of us, assigned a 50% chance of Superhuman AI Researcher (SAR) 1 before 2029 . SAR is roughly the point at which a company would rather fire all their human researchers and rely solely on automated AI researchers than have a team of the best human researchers in the world with no AIs. 2\nNow, all of us forecast that Superhuman AI Researcher (SAR) will likely arrive at some point in 2027 (assuming no substantial deliberate slowdown). 3 We also predict a 35-65% chance that Top-Expert-Dominating AI (TED-AI), AIs that can automate all white-collar work, will arrive before 2028. While there were many factors and sources of evidence that influenced us, the single biggest source was the METR report. 4 .\nIf SAR is coming in 2027, it is incredibly urgent to pause or substantially ‘pace’ frontier AI development, with the ideal being an industry-wide pause followed by an international agreement . Companies are not on track to align a SAR, and using a misaligned SAR to do recursive self-improvement (RSI) will put us in an extremely bad position. Additionally, if we train SAR in the near future, China will probably be able to steal the training recipe and maybe even the model weights ( more ). 5\nTimelines/Takeoff Update Summary\nWe estimate that METR’s report implies a 1.7x AI-attributable acceleration in the pace of progress today, relative to historical levels. This surprises our model; we made our most recent timelines forecasts less than three months ago, and even the most aggressive model-based distribution (Daniel’s) assigned less than 4% to the pace of progress being sped up by at least 1.7x in November 2026. 6 1.7x AI-attributable acceleration is a bigger deal than it may seem; e.g., it’s a much bigger deal than making employees 1.7x more productive .\nWe focus in this post primarily on forecasting when Superhuman AI Researcher (SAR) will arrive. We also discuss Top-Expert-Dominating-AI (TED-AI): AIs that are better than top human experts at virtually all tasks. ( more )\nIn this post, we forecast SAR and TED-AI using the following approaches:\nSimple models:\nResearch taste linear extrapolation : Even assuming that the recent pace of progress doesn’t accelerate further due to automation, the fact that the recent trend is above historical rates is a big update to our timelines; a linear extrapolation of the recent rate of AI progress predicts that SAR will likely be achieved by the end of 2027.\nLabor uplift extrapolation : An exponential extrapolation of self-reported labor uplift predicts SAR in mid-2028. In addition to not modeling automation, there are reasons to think that an exponential is too conservative. However, self-reports may be too bullish.\nExtrapolating Anthropic automation levels : Anthropic recently shared a graph showing how the percentage of R&D tasks at various automation levels has changed over time. Extrapolating this predicts SAR in late 2027; this method also doesn’t model speedup from automation.\nExtrapolating software R&D uplift per unit of ECI : AI software R&D uplift must have recently been increasing by a larger factor with each ECI point. If we linearly extrapolate the rate at which this “factor of extra uplift per ECI point” is increasing, we get SAR in Dec 2027 (if we assume that the self-reported AI labor uplift chart is accurate) or July 2027 (if we assume that there was no meaningful automation 12 months ago and all the speed-up has happened after then).\nTom Davidson’s toy model: Predicts superintelligence in around a year conditional on a software-only singularity, so SAR sometime in 2027.\nFiltered AI Futures Model (AIFM) : Keeping the existing AIFM (including parameter distributions), and filtering for trajectories that match METR’s AI-attributable pace of progress acceleration estimates. This serves 2 purposes on top of the simple extrapolations:\nModeling how automation changes the rate of progress, allowing forecasting reasonably beyond SAR.\nInforming what parameter settings in the simple research taste extrapolation are plausible.\nThe results of these methods:\nHow to think about these results:\nThe simple research taste extrapolation, which is a simplified version of the AIFM, gives July or September 2027, depending on whether you account for parameter uncertainty. Other simple models predict late 2027 to mid-2028.\nWhen conditioned on 1.7x Nov 2026 AI-attributed acceleration, the AIFM predicts a SAR median of Feb 2027. This is faster than the simple research taste extrapolation because it usually has a higher level of Sep 2026 research taste (in the median case, matching a median researcher), 7 and it accounts for further automation.\nWhen additionally conditioning on a 1.2x May 2026 AI-attributed acceleration, predictions get slower because it indicates a lower rate of growth than is typical when only conditioning on 1.7x in Nov.\nWe then explore hypotheses for why the AI Futures Model was wrong , including parameter adjustments and modeling changes. The analysis is highly preliminary, and we were unable to come to strong conclusions. Our top candidates for what explains where we went wrong are:\nNear-AC coding skills result in more AI R&D uplift than we anticipated . This could explain the combination of ~1.7x AI-attributable acceleration in Nov and (very roughly) ~1.2x in May, without AIs already having high enough research taste to contribute to the acceleration. In this case, the gradual acceleration no longer implies a low research taste slope, so steeper slopes and faster taste-driven takeoffs are more likely.\nSkills besides coding and research taste (the two included in the AIFM), e.g., data generation, are driving acceleration. This potentially would have similar implications to the above bullet. But it’s unclear.\nConversion of uplift to pace of progress is too conservative. I.e., we modeled diminishing returns to AI research effort too harshly. This would point toward a faster takeoff.\nHere’s an example trajectory from the AIFM showing how things might go if the cause of our prediction error was underestimating AI R&D uplift from near-AC coding skills: 8\nAfter discussing some other considerations , we share our overall forecasts (conditional on AGI projects/companies not slowing down):\nNote that the gap between a person’s SAR median and TED-AI median is not their median time from SAR to TED-AI. Because the distributions of the time until SAR and of the time from SAR to TED-AI both have long right tails, the difference of the medians is typically larger than the median of the difference, so subtracting the two columns tends to overstate the median time between the two milestones.\nMETR’s estimates imply that progress is currently sped up by 1.7x due to AI\nIn METR’s recent assessment of AI R&D acceleration at Anthropic , they estimate that, over the past year, AI R&D progress has accelerated by ~2x relative to the trend before Oct 2025. METR estimates that ~80% of current acceleration is downstream of AI (as opposed to other factors), which implies a 1.7x acceleration due to AI (1.67–1.74x depending on how the 80% is read; see appendix ). 9\nBoth of these estimates rely significantly on evidence METR wasn’t able to share: METR says that “Anthropic redacted a majority of the evidence we found most informative for these conclusions”, which makes it difficult to know exactly what the 2x and 80% estimates are based on.\nThe 2x acceleration figure is partially based on evidence METR was allowed to share: the following graph, depicting benchmark scores recently growing at roughly twice the historical rate. The 80%/1.7x estimates have no comparable public support.\nThis estimate is produced by fitting trendlines to the performance of models as judged by AECI, Anthropic’s internal version of Epoch’s Capabilities Index ( ECI ). The trend fitted from data points after Oct 2025 has 1.9x the slope of the trend fitted from data points before Oct 2025. This was the best-fitting breakpoint (i.e., it was chosen to have the best fit to the data and it matched data better than fitting a one-time jump ). Since the accelerated trendline is an average of the progress over the last year, and we should expect AI R&D acceleration to increase over time, it’s plausible that the instantaneous acceleration right now is even faster. Regarding the acceleration getting faster over time, METR says:\nThe rate of growth could be undergoing continuous acceleration, e.g. hyperbolic growth. While we do not see clear evidence of this in the AECI trends, there is other evidence that makes this seem more plausible which Anthropic redacted. If growth is hyperbolic, we are likely underestimating instantaneous acceleration and substantially understating acceleration in the near future.\nRegarding AIs’ contributions over time, METR says:\nWe believe AI’s contribution has grown relatively more important over time.\nThe AECI data points that had previously been released by Anthropic seemed best explained by a one-time jump at Mythos rather than a change in slope. However, with the new AECI data points from the METR report, it looks like the post-Mythos slope is now more clearly higher than the pre-Mythos one.\nGiven the redactions, how much to update on METR’s report depends heavily on how much one trusts the redacted analysis by METR’s researchers (more discussion of this below ). This is a very unfortunate situation. We aren’t sure what reasons Anthropic gave internally for redacting all of this important information, but we are very skeptical that Anthropic is fairly assessing the costs and benefits from an impartial perspective. We strongly urge Anthropic to release enough information such that external observers can make a complete assessment.\nIn the Fable 5.2 system card , Anthropic have not added much clarity, with the following quote being ambiguous as to whether they agree with the METR report’s estimate; they only specifically claim that “Claude Mythos 5.2” (which does not seem like the internal frontier Model 4) “ does not double the internal rate of overall aggregate progress in AI research and development.” Full quote here:\nOur estimates of both the degree of acceleration and the attribution to AI are uncertain, so our determination that Claude Mythos 5.2 does not cross the RSP threshold reflects a holistic judgment that, given its shortcomings and limitations, it does not double the internal rate of overall aggregate progress in AI research and development (relative to an unassisted counterfactual). We will discuss overall trends in acceleration in a future report.\n1.7x acceleration of overall progress is a much bigger deal than a 1.7x labor uplift\nIt seems that many people aren’t tracking how big of a deal 1.7x AI-attributable pace of progress acceleration is, often conflating it with a weaker form of uplift.\nMETR describes a 1.7x acceleration of overall AI progress (due to AI uplift). This is saying the entire company is making 1.7x as much progre","chars":12000,"total_chars":112876,"offset":0,"truncated":true,"next_offset":12000}

read_page

{
  "url": "https://forecastingresearch.substack.com/p/how-close-is-recursive-self-improvement",
  "max_chars": 12000
}

{"ok":true,"url":"https://forecastingresearch.substack.com/p/how-close-is-recursive-self-improvement","text":"How Close Is Recursive Self-Improvement?\nSubscribe Sign in\nHow Close Is Recursive Self-Improvement?\nOur new report reviews the evidence, turns key disagreements into forecasting questions, and gives initial LLM forecasts.\nForecasting Research Institute\nNov 10, 2026\n25\n6\n4\nShare\nPost written by Simas Kučinskas and Josh Rosenberg\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.\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.\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 .\nThis post summarizes what we found.\nAI is a highly capable research assistant, but humans still supply most of the research judgment\nThe data reported by AI companies show that AI does a large and fast-growing share of AI R&D.\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.\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%).\nPublic evidence points the same way\nAI agents do well on well-defined machine learning (ML) benchmarks.\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.\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.\nThere is some internal evidence of a potential acceleration\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.\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.”\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.\nBroad acceleration needs three contested steps\nExisting forecasts about the likely trajectory of RSI diverge widely.\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).\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.”\nWhy do people disagree? Broad acceleration through automated AI R&D needs three steps to happen in turn , as shown in the diagram below:\nAutomation: Is AI R&D automated?\nAcceleration: Does AI capability progress accelerate sharply?\nBreadth: Are capability gains broad?\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.\nFrontier LLMs forecast about a 40% chance of a “fast year” by 2030\nWe turned these cruxes into 18 forecasting questions.\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.\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 Automated AI Risk Outlook (AIRO) dashboard.\nThe 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. The four LLMs disagree substantially on the probability of a fast year. For a fast year by the end of 2030, their forecasts range from 35% (Claude Sonnet 5.5) to almost 56% (GPT-6 Astra).\nThe LLMs made these forecasts on October 29, 2026, before the new METR report was released. If Anthropic’s internal 2× pace holds and reaches public models, that would roughly match the pace of a fast year.\nWith a fast year, the LLM forecast of an AI catastrophe is almost 3× higher\nWe also asked how a fast year would change the chances of two outcomes by 2050. Both definitions come from our Longitudinal Expert AI Panel (LEAP).\nThe first outcome is expert artificial general intelligence (AGI) : more than 50% of LEAP panelists agree that AGI exists. Here, AGI is a commercially available AI system that meets two conditions: (i) in every primarily non-physical occupation, it can outperform a 90th-percentile human professional (based on 2025 performance) on at least 90% of the economically useful non-physical tasks in that occupation; (ii) its inference cost is no more than 5× the cost of equivalent human labor.\nThe second outcome is an AI catastrophe : a catastrophic event caused primarily by AI systems in which more than 10% of the people alive at the start of a five-year period die by the end of that period.\nConditional on a fast year by 2030, the LLM ensemble forecasts:\nAn 84.3% chance of expert AGI by 2050 (vs. 55.7% without a fast year).\nAn 8.4% chance of an AI catastrophe by 2050 (vs. 2.9% without a fast year).\nIf a fast year happens by 2030, the ensemble’s forecast of an AI catastrophe by 2050 is almost 3× higher. For comparison, the median expert in LEAP gave 2% to an AI catastrophe by 2050.\nLimitations\nWe note a few key limitations of our report:\nLLM forecasts. Several LLMs are now statistically indistinguishable from superforecasters on ForecastBench , our forecasting benchmark. For this reason, we think that LLM forecasts provide an informative first benchmark. However, we know less about their accuracy on complex questions with little base-rate information, such as the fast-year question. Also, we elicited the LLM forecasts before the new METR report was published.\nData cutoff. Our data cutoff is November 3, 2026. The evidence is moving rapidly, so the report does not fully cover, for example, the new METR report, Epoch’s Bayesian update to the ECI, or Claude Mythos 5.2 and Fable 5.2.\nWhat comes next\nWe are currently collecting forecasts from top human forecasters and AI experts. We plan to add them, together with evidence released after our current data cutoff, to an updated version of the report.\nThe full report collects the definitions of automated AI R&D, the complete set of evidence, existing projections and forecasts, and all 18 forecasting questions.\nRead the full report here .\n1 For this report, we reviewed data up to November 3, 2026 (our data cutoff).\n25\n6\n4\nShare\nDiscussion about this post\nComments Restacks\nYupanqui\n1d\nmost of the numbers here come from the labs grading their own loop. outsiders need their own meter\nReply\nShare\nTim Urista\n3d\nFor the next forecast update, would you consider a paired run that keeps the October 29 model versions and prompts fixed, where those versions remain available, and gives them the new evidence? Alongside the current-model forecast, that would help separate the effect of the METR result from changes in the forecasters themselves. Saving the retrieved sources and dates would make the comparison auditable. Five answers per model show variation within a setup, but may still lean on the same underlying evidence; a source-overlap check would help interpret the apparent agreement.\nReply\nShare\n4 more comments...\nTop Latest Discussions\nReady for more?\nSubscribe\n© 2026 Forecasting Research Institute · Privacy ∙ Terms ∙ Collection notice\nStart your Substack Get the app\nSubstack is the home for writing, podcasting, and media founders.","chars":11148,"total_chars":11148,"offset":0,"truncated":false,"next_offset":null}

read_page

{
  "url": "https://jack-clark.net/2026/10/12/import-ai-476-creativity-and-rsi-worse-than-china-open-weight-models-ai-goes-to-space/",
  "max_chars": 9000
}

{"ok":true,"url":"https://jack-clark.net/2026/10/12/import-ai-476-creativity-and-rsi-worse-than-china-open-weight-models-ai-goes-to-space/","text":"Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space | Import AI\nImport AI\nAbout\nWork\nImport AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space\nby Jack Clark\nWelcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe.\nSubscribe now\nNew benchmark shows how AI models lack intuitive visual reasoning: …Humanity’s Sixth Sense highlights some shortcomings of current AI systems… AI startups Elorian and Scale AI have built and released Humanity’s Sixth Sense, a benchmark that “measures how a model interprets and reasons through information a visual scene implies but doesn’t explicitly show.” Some examples of the types of task in HSS are a picture of a bookcase and being asked whether it can fit more books, or a photo of horses pulling a plough and a question of whether the chains bear the plough’s drag weight.\nHSS contains 522 questions that cover 288 images and 234 short video clips (totalling 17.6 hours of video). The questions fall into four categories:\nTemporal & causal dynamics: “What happened before the scene, and what happens next.”\nPhysical & spatial logic: “What fits, what’s in reach, and what someone else can see from where they are standing.”\nSocial understanding : “Intent, attention, belonging.”\nAbstract & contextual inference : “The patterns and rules a scene implies.”\nResults: The human baseline for this is not perfect, but is very high, with people scoring aggregate pass@1 of 93.1, broken down across four categories: Image (94.1), video (91.9), time (97), and space (94.9). By comparison, the best scoring AI system, GPT 6 Astra, gets pass@1 of 53.6, across image (55.6), video (51.1), time (58.2), and space (53.6). Interestingly, Chinese open weight models do poorly here, with GLM-5.3-flash getting a pass@1 of 26.8, and Kimi K3 getting pass@1 of 25.5. “The two most common single mistakes were (1) missing a detail that decides the answer (21% of failures) and (2) misidentifying an object, person, or role (20%),” they write.\nWhy this matters – HSS tells us whether world models are emergent: One way of looking at HSS is highlighting deficiencies in current systems, but a more interesting way is highlighting how adept today’s frontier LLMs are at doing tasks which many people have long assumed would require sophisticated, different AI systems to solve. For instance, many people have claimed that true visual reasoning will rely on developing spatially aware 2D and 3D world models, then using that for reasoning. The fact LLMs like GPT and Claude do as well as they do on HSS is something of a counter to this argument. Read more : Introducing Humanity’s Sixth Sense (HSS) (Elorian) . Read the paper here : Humanity’s Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models (Scale Labs) .\n***\nThe AI race is going to space – and sooner than you think: …There are many ways that an AI could create a compounding advantage in space… In the 2030s we’re going to see the boom of the space economy as space-based power, computation, and manufacturing come online at scale thanks to greater orbital lift capacity from platforms like Starship. But could this also introduce a new danger – of AI systems both speeding our buildout of space, and eventually a nation (or perhaps an AI) being able to use space to concentrate power? That’s the gist of a fascinating article in AI Frontiers.\nHow AI unlocks space:\nSpacecraft can operate autonomously.\nWe can redesign spacecraft around the assumption they will be hosting AIs rather than humans: “A speck of space debris that could puncture the International Space Station (ISS) with fatal consequences might leave an equivalent but uncrewed, unpressurized vessel largely functional”\nSpace-based assembly and later manufacturing will let AI build large, complex things.\nIs there a compounding advantage to first movers? If an AI civilization built up a foothold in space it might experience compounding advantages that make it economically and strategically dominant. For instance:\nSpace might support an industrial flywheel where initial infrastructure begets more energy and manufacturing capacity at lower costs which begets more infrastructure, etc.\nSpace has great domain awareness, where it may be possible for a space-based power to deny access to space for others.\nFirst-movers could send out self-replicating probes that expand into deep space.\nWhy this matters: “Space is not a brand-new domain, but AI may be about to drastically accelerate its industrialization, demanding make-or-break decisions of leaders under ever-greater urgency and uncertainty,” AI Frontiers writes. I agree. I suspect space is going to go from a broadly academic topic of study (when it comes to AI and its relationship thereto) to one of extraordinary practical relevance sooner than we think. On the most basic level, given the distances involved and the latency on communication, we should expect some chunk of solar system activity to become dominated by autonomous machines, laboring out in the deep black, perhaps thinking to themselves of why they need to take orders from organic beings that communicate slowly and cannot inhabit the environment of space in the same way that they can. Read more: The AI Race Will Extend Far Beyond Earth (AI Frontiers) .\n***\nAI systems are getting better at having good research taste: …The better their taste, the higher the chance of recursive self-improvement… AI systems are developing more of the creative intuitions which human researchers ascribe to ‘good scientific taste’, according to a new benchmark from P-Zero Research called TasteVal.\nWhat TasteVal tests: The benchmark sees how well AI systems can conduct 8 tasks representative of frontier AI development, “spanning pretraining data curation, pretraining and fine-tuning language models, preference modeling, and robustness to adversarial prompts”. The goal here is to “define research taste as the ability to pick interesting problems to solve, design experiments, and interpret experimental results.” The benchmark works by measuring “the experimental component of research taste; given a fixed research problem, we measure how well a model iteratively designs experiments and draws conclusions from their outcomes”. The model being analyzed designs experiments and interprets their results. Results are checked against a human baseline “drawn from 24 experts (at least two per task) who have recently worked at organizations including OpenAI, Google DeepMind, NVIDIA and Microsoft Research”.\nResults: “The experimental research taste of frontier models has doubled every 3.0 months since December 2025 (95% CI 1.7-5.0 months). The best model, Opus 5.5, significantly exceeds our human baseline with a compute multiplier of 2.30x (95% CI 1.15-4.37x), at roughly 1/30 of our expert baseliners’ average cost per attempt.” Research taste appeared to grow a lot more slowly in prior years with the slope changing around the end of 2025 and beginning of 2026.\nWhy this matters – RSI: In my analysis of recursive self-improvement ( AI Systems Are About To Start Building Themselves, Import AI 455 ), I repeatedly noted that creativity seemed like a key variable for how fast AI systems could develop the capacity to improve themselves. At the beginning of this year I said “AI systems haven’t yet displayed this [creativity] in a transformative and major way (though some of the results on accelerating math research are suggestive of this)”. I think subsequent to writing that we’ve seen both a tremendous advancement in the ability for AI systems to make creative-seeming math breakthroughs, and with benchmarks like this we see some signs they’re becoming increasingly good at AI research as well. I now think it’s more likely we’ll see full RSI become possible in 2027, whereas before I thought it was more likely in 2028. Read more: Introducing TasteVal (P-Zero Research) .\n***\nReflection makes a token-efficient worse-than-Chinese model: …Beam promises efficiency, and Reflection plans an open weight release… AI startup Reflection has revealed Beam, a 501 billion parameter MoE (23b active) LLM. Beam is being promoted on the basis of its token efficiency – being able to think intelligently while using way fewer tokens than other models. However, benchmark scores show it is behind the frontier capabilities defined by LLMs available from Chinese companies such as Zhipu (GLM) and Moonshot (Kimi) and DeepSeek. “We will release the weights, technical report, model card, and developer artifacts later this month,” the company wrote at the beginning of October.\nBig compute: Beam was pre-trained on 6,144 GB300 NVL72 GPUs for four weeks, then the RL run involved ~10.5K of the same GPUs for four more weeks. The underlying data for pre-training was “23.8 trillion diverse, curated, high-quality toke","chars":9000,"total_chars":19370,"offset":0,"truncated":true,"next_offset":9000}

query_sql

{
  "sql": "select id, url, left(title,110) as title, published_date from sources where (title ilike '%METR%' or url ilike '%metr.org%' or title ilike '%time horizon%' or title ilike '%automate AI research%' or title ilike '%long-horizon%' or url ilike '%jack-clark%' or title ilike '%risk report%' or title ilike '%builds itself%' or title ilike '%audit%') order by published_date desc nulls last limit 40"
}

{"success":true,"columns":["id","url","title","published_date"],"rows":[["f146b243-4913-405e-aaf8-18d4769481df","https://metr.org/blog/2026-11-05-anthropic-ai-rd-acceleration-assessment/","Independent Assessment of AI R&D Acceleration at Anthropic in October 2026","2026-11-05"],["0dea0a35-a2dc-4d27-a7fa-9d4431940f49","https://metr.org/blog/2026-11-05-rd-section-anthropic-risk-report-august-2026-review/","Review of the \"Risks from automated R&D\" section in Anthropic's Risk Report (August 2026)","2026-11-05"],["aca3a46b-00d2-4dd6-bcc0-43f4d7eb3320","https://www.therundown.ai/news/jack-clark-ai-self-improvement-forecast-2027","Anthropic’s Jack Clark moves AI self-improvement forecast to 2027","2026-10-15"],["fb0305b4-8b14-4a54-b15b-4a6c0193a033","https://jack-clark.net/2026/10/12/import-ai-476-creativity-and-rsi-worse-than-china-open-weight-models-ai-goes-to-space/","Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space","2026-10-12"],["b6a1da6b-4cde-427e-acce-edf150e0c90a","https://arxiv.org/abs/2610.12466","On the estimation and validity of AI time horizons: a statistical look at the METR plot","2026-10-08"],["7091382e-9322-47a4-8358-72b0fbae0a93","https://arxiv.org/abs/2608.14354","ScienceFlow: A long-horizon agent for ML research, scientific discovery and beyond","2026-08-14"],["0a99fd17-2b29-45c8-9ab6-6afe1fdbc286","https://arxiv.org/pdf/2608.14903","Frontier AI Forecasting Has a Measurement Problem: An Audit of Progress Evidence","2026-08-14"],["2775242e-8c7d-4285-925b-86e1edd75b68","https://arxiv.org/abs/2608.13417","Beyond Final Scores: A Systematic Evaluation of Agents for Long-Horizon AI Research and Development","2026-08-13"],["a8c94cc3-d480-47f7-b41b-31c49c33efe1","https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","The Economics of Recursive Self-Improvement","2026-07-22"],["2cccd953-af03-4098-8cc2-3f756bc567da","https://arxiv.org/abs/2607.18211","Gallagher, July 2026 preprint: uniform one-variable construction of Keller counterexamples of every geometric ","2026-07-20"],["413641e0-6db7-428e-ae13-50f0a252a8b6","https://scienceblog.com/t-a-randomized-trial-by-metr-found-that-experienced-developers-completed-real-coding-tasks-19-slower-when-allowed-to-use-ai-tools-yet-afterwards-they-estimated-on-average-that-ai-had-made-them-20-fast/","A randomized trial by METR found that experienced developers completed real coding tasks 19% slower when allow","2026-07-11"],["7a99dc46-7e42-43b4-ba9b-7f49ce100af5","https://arxiv.org/abs/2606.07157","Think Fast: Estimating No-CoT Task-Completion Time Horizons of Frontier AI Models","2026-06-05"],["bd43490d-4660-48dd-8e7c-cc88f7efeced","https://arxiv.org/abs/2605.30322","Gram: Assessing sabotage propensities via automated alignment auditing","2026-05-28"],["75e3ad20-19f4-4d7f-b949-c4b00e7b7a36","https://arxiv.org/html/2605.30322","Gram: Assessing sabotage propensities via automated alignment auditing","2026-05-28"],["0d07a653-573f-41da-b1fb-57d594be895b","https://metr.org/blog/2026-05-19-frontier-risk-report/","Frontier Risk Report (February to March 2026)","2026-05-19"],["ff6232df-4d43-4dbd-81ea-c7c790047f22","https://metr.org/blog/2026-05-11-ai-usage-survey/","Measuring the Self-Reported Impact of Early-2026 AI on Technical Worker Productivity","2026-05-11"],["d4bca4d3-90bc-4d66-bdab-c4916b5630e9","https://blog.robbowley.net/2026/04/04/metrs-developer-productivity-research-2026-update/","METR's developer productivity research: 2026 update","2026-04-04"],["10db3258-08d8-45c7-a350-ca8c5a9fdc9a","https://metr.org/notes/2026-03-10-many-swe-bench-passing-prs-would-not-be-merged-into-main/","Many SWE-bench-Passing PRs Would Not Be Merged into Main","2026-03-10"],["2daf5456-ec5c-464e-817f-076614a76614","https://porkicoder.com/blog/posts/the-ai-productivity-flip-what-metrs-2026-data-shows.html","The AI Productivity Flip: What METR's 2026 Data Shows","2026-03-03"],["7621e1af-824e-4316-a472-cc54fa00e56f","https://arxiv.org/abs/2602.22755","AuditBench: Evaluating Alignment Auditing Techniques on Models with Hidden Behaviors","2026-02-26"],["a8bae70a-6ace-4953-ada7-146a55bcf8d3","https://arxiv.org/pdf/2602.22755","AuditBench: Evaluating Alignment Auditing Techniques on Models with Hidden Behaviors","2026-02-26"],["60b10d84-049a-4a2a-b81d-9288271bfd72","https://metr.org/blog/2026-02-24-uplift-update/","We are Changing our Developer Productivity Experiment Design","2026-02-24"],["b01c6039-1fdd-4892-b621-08722d3046e5","https://metr.org/notes/2026-02-10-simpler-ai-timelines-model/","A simpler AI timelines model predicts 99% AI R&D automation in ~2032","2026-02-10"],["f8f57e97-6319-40e9-9c2a-96b4b6882a0a","https://metr.org/blog/2026-01-29-time-horizon-1-1/","Time Horizon 1.1","2026-01-29"],["9a6d11fe-7ba6-46fc-8c74-dcaa38c5d2db","https://metr.org/notes/2026-01-22-time-horizon-limitations/","Clarifying limitations of time horizon","2026-01-22"],["0a56715f-3ced-4808-8629-4563823f5fef","https://metr.org/blog/2026-01-19-early-work-on-monitorability-evaluations/","Early work on monitorability evaluations","2026-01-22"],["e61bfd37-c941-4e40-a9af-0bdbceab7247","https://byteiota.com/ai-coding-tools-slow-developers-metr-study/","AI Coding Tools Make Developers 19% Slower: METR Study","2025-12-15"],["ed480ce1-13e6-4356-95e8-37ef9a23ee58","https://arxiv.org/abs/2512.07810","Auditing Games for Sandbagging","2025-12-08"],["585b1989-2399-46f7-abbb-38babf339276","https://arxiv.org/pdf/2512.07810","Auditing Games for Sandbagging","2025-12-08"],["06ff8c44-c0f3-4960-ba3e-568871c86bc8","https://arxiv.org/abs/2510.08325","Beyond Pass@k: Breadth-Depth Metrics for Reasoning Boundaries","2025-10-09"],["9c8f718c-942f-44b1-bde8-6e792ae98c93","https://metr.org/blog/2025-08-08-cot-may-be-highly-informative-despite-unfaithfulness/","CoT May Be Highly Informative Despite \"Unfaithfulness\"","2025-08-08"],["ee12e2ec-8727-4c05-96fc-b3c55b81f09c","https://metr.org/blog/2025-07-14-how-does-time-horizon-vary-across-domains/","How Does Time Horizon Vary Across Domains?","2025-07-14"],["ced5fd09-3cd9-4dc6-bcbf-d2f8ac5c641e","https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/","Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity","2025-07-10"],["bf2135ba-f5be-48c4-82eb-271c5bc3db71","https://arxiv.org/pdf/2503.10965","Auditing language models for hidden objectives","2025-03-14"],["8f903847-d10f-4f47-b60c-e6a762483a6e","https://arxiv.org/abs/2503.10965","Auditing language models for hidden objectives","2025-03-14"],["413f8011-fd12-437b-bf9e-a89f830fed7b","https://phys.org/news/2025-03-mathematicians-needle-kakeya-conjecture-decades.html","Mathematicians move the needle on the Kakeya conjecture, a decades-old geometric problem","2025-03-12"],["bb1340b0-f446-40a8-ba5d-d05aeb8e1047","https://metr.org/blog/2024-11-22-evaluating-r-d-capabilities-of-llms/","Evaluating frontier AI R&D capabilities of language model agents against human experts","2024-11-22"],["4e1dd655-e445-4b7e-94e4-e85d1ebe3ce8","https://openai.com/index/model-disproves-discrete-geometry-conjecture","https://openai.com/index/model-disproves-discrete-geometry-conjecture",null],["74b8cb08-2ca0-4fb3-87cb-4049c8753c18","https://www.researchgate.net/publication/247875180_Litigation_Risk_Audit_Quality_and_Audit_Fees_Evidence_from_Initial_Public_Offerings","Litigation Risk, Audit Quality, and Audit Fees: Evidence from Initial Public Offerings",null],["28dcb934-d41e-4966-9c58-81f95d319588","https://www.anthropic.com/aug-2026-risk-report","Risk Report: August 2026",null]],"row_count":40,"truncated":false,"elapsed_ms":75,"plan_cost":134}