A new arXiv paper investigates how giving large language models access to external tools they do not actually need changes the way they answer questions. The authors find that mere availability of tools can shift model behavior even when no external information is required. The work suggests tool provisioning should be matched to the task rather than offered by default.
A new arXiv paper argues that causal claims about the internal workings of large language models depend on measurements such as projections, cosine similarities, ablation deltas, and interchange patches. The authors catalog the specific ways these instruments fail and propose calibration steps to take before relying on their results. The work is listed under arXiv's machine learning and AI categories.
A revised arXiv paper examines why exploration helps in reinforcement learning, arguing it only pays off when agents revisit similar states repeatedly. The authors show that without such retries, a purely greedy policy would be optimal, and study how exploration behavior can emerge in policy gradient methods.
A new arXiv preprint examines the kinds of mistakes large language models make while working through mathematics problems, grouping them into distinct error categories. The authors use that taxonomy of reasoning failures to target improvements in the models' mathematical problem-solving. The work aims to give a clearer picture of where current LLM reasoning breaks down and how to address it.
A new arXiv preprint proposes "Discovery Foundation Models," framing open-ended discovery as the next stage for AI systems. The authors argue that models have moved from recalling and reasoning over existing knowledge to acting with tools and learning from outcomes, and that the next step is generating genuinely new findings. The paper is a position piece rather than an experimental release.
A revised arXiv preprint (2609.06396v2) introduces MetaRSI, a framework that applies recursive self-improvement at a meta level, treating the model-building machinery itself as the target of improvement so later generations inherit the gains. The authors argue that RSI research has so far been validated almost exclusively on coding and formal benchmarks such as science QA, and their approach seeks to broaden where such gains hold.
A widely discussed essay contends that progress in artificial intelligence is hampered by how the field identifies and validates new discoveries. The piece argues that current research incentives and evaluation practices make it harder to distinguish genuine advances from incremental or overstated results.
OpenAI has published an inside look at how coding agents are changing the way its teams conduct AI research. The post presents early figures on agent adoption, the speed of experiments, and the complexity of tasks being handed off to AI systems. The findings suggest these tools are streamlining research workflows across the company.
WHY IT MATTERS ↘First-party metrics from a frontier lab give practitioners a rare, if self-reported, benchmark for how much coding agents actually compress research cycles, informing adoption and cost decisions. They also signal that AI-accelerated R&D is becoming a compounding competitive advantage among labs.