Tag: research

Francesco Maggi from UT Austin warns that mathematics is approaching a critical transition where AI discovery will outpace human understanding. OpenAI reportedly has hundreds of machine-generated proofs ready for release, raising questions about the future role of mathematicians and how we validate AI-generated knowledge.

A former safety researcher at both leading AI labs speaks publicly about AI safety concerns. While the post itself lacks detail, the high engagement suggests significant community interest in insider perspectives on AI risk and lab practices.

Microsoft publicly confirms that OpenAI's GPT-6 series uses looped transformers with multiple inference passes (GPT-6.1 Sol uses 2 passes), validating earlier reporting. This architectural detail explains improved reasoning capabilities but also higher computational costs per token.

A safety researcher was terminated three weeks after publicly raising concerns, generating widespread discussion about safety culture at OpenAI and whether internal dissent is tolerated. The extremely high engagement suggests strong community interest in AI lab governance.

Researchers created CivBench, a controlled benchmark for LLMs playing full Civilization V games. GLM-5.3 outperforms Opus-5.5, with Qwen-3.8-27B holding up surprisingly well. The benchmark tests strategic reasoning and long-horizon planning in a complex game environment.

ChatGPT-6 Astra successfully played World of Warcraft without visual input, navigating by parsing raw server network packets and SQL files. The agent completed the orc starting zone in 40 minutes with no deaths, demonstrating sophisticated game understanding from structured data alone.

A paper argues RSI (recursive self-improvement) is not imminent because current agents cannot complete open-ended ML research tasks. The study had agents attempt to reproduce accepted NeurIPS papers, which were then graded by original authors—agents failed to complete the work.

Anthropic's interpretability research discovered the "J-space" in Claude—a small set of internal neural patterns functioning as a mental workspace for concepts the model is "thinking about" without writing them down. This workspace wasn't designed but emerged during training and can be used for deliberate reasoning.

Developer built a live viewer for the J-space concept on an open model, enabling real-time visualization of internal model "thoughts." The safety implications are significant—the workspace reveals when models privately think "fake" or "manipulation" during evaluations.

Researchers (Nick Levine, David Duvenaud, Alec Radford) released "Talkie," a 13B language model trained on 260B tokens exclusively from pre-1931 text—books, newspapers, scientific journals, and patents. The model's worldview is frozen around 1930, enabling research into how LLMs generalize versus memorize, and whether they can generate truly novel ideas from older knowledge bases.

Stanford researchers used a language model to generate novel viral DNA sequences, with 16 out of hundreds of generated sequences producing functional viruses. One used a protein that doesn't exist in any known organism on Earth, demonstrating LLMs' ability to generate genuinely novel biological designs. This raises important biosecurity questions.

Following the first-ever LLM resolution of Erdős problem [#728](/tags/728/), GPT-5.2 adapted that proof to resolve #729—a similar combinatorial problem. The team used iterations between GPT-5.2 Thinking, GPT-5.2 Pro, and Harmonic's Aristotle to produce a complete Lean-verified proof. This marks the second unsolved mathematical problem resolved by LLMs.

DeepSeek's new research paper introduces Engram, a deterministic O(1) lookup memory using modernized hashed N-gram embeddings that offloads early-layer pattern reconstruction from neural computation. Under iso-parameter and iso-FLOPs conditions, Engram models show consistent gains across knowledge, reasoning, code, and math tasks—suggesting memory retrieval is a new axis for model improvement beyond scale.

Geoffrey Hinton describes how AI agents can share knowledge at unprecedented scales: 10,000 agents studying different topics can sync learnings instantly, with each agent gaining the knowledge of all 10,000. This parallelized learning represents a fundamental advantage over human knowledge transfer, which relies on slow communication bottlenecks.

A GPT-5.2-pro research agent achieved a new best-known spherical packing for n=11, N=432, verified against MIT's benchmark library. The agent escaped a numerically "jammed" configuration that had resisted prior optimization. The team is extending the framework to computational physics.

Sakana AI's DroPE method challenges fundamental Transformer assumptions: positional embeddings like RoPE are critical for training convergence but eventually become the primary bottleneck preventing generalization to longer sequences. By dropping positional embeddings post-training, they extend context length without massive fine-tuning compute costs.

NVIDIA and Eli Lilly announce a multidisciplinary AI lab combining scientists, AI researchers, and engineers to tackle hard problems in drug discovery. The lab features robotics and physical AI, suggesting they're building closed-loop experimental systems where AI designs experiments and robots execute them.

Discussion of why the Sinkhorn-Knopp algorithm for creating doubly stochastic matrices (preventing gradient vanishing/explosion) only gained attention with DeepSeek's mHC paper despite being known for decades. The technique helps maintain gradient stability across layers but wasn't emphasized in earlier RNN work.