Claims Browser

Biological-Hybrid Intelligence systems require reciprocal co-adaptation between biological and artificial substrates and differ from systems that merely decode biological activity or stimulate living substrates

"BHI requires reciprocal co-adaptation and differs from systems that merely decode biological activity, stimulate a living substrate, or adapt a single component."

BHI reframes biological-artificial integration as a system-level problem of computational allocation, coordination, and control

"BHI therefore reframes biological-artificial integration as a system-level problem of computational allocation, coordination, and control."

Training with debate where two AIs argue can mitigate reward hacking when using LLM judges for RL training

"We show that training with debate, where two AIs argue against each to convince a judge, can mitigate reward hacking, potentially providing a hopeful direction for scaling up accurate training supervision for fuzzy tasks."

Debate training recovers about 45% of the gap between LLM judge training and ground truth training in terms of peak accuracy

"Debate recovers about 45% of the gap between the peak accuracy of training a single policy with an LLM judge and the peak accuracy of training with the ground truth answers."

Future RL training will incorporate current-generation AI systems to provide reward signals for next-generation AI

"This is our current best-guess model of future AI development, and so the role of debate is to ensure that the LLM judgements used during training provide as accurate a reward signal as possible."

LLM judgments will become an increasingly important aspect of frontier RL training, especially for long agentic trajectories

"The trend of increasing test time compute will likely only exacerbate this problem. Even for tasks like coding, there are many aspects of desirable LLM agent behavior that are fuzzy, and so LLM judgements are likely to become an increasingly important aspect of frontier RL training."

A graphical notation adapted from Penrose tensor notation can provide a global view of interpretable AI architectures and map directly to PyTorch einsum code

"This graphical notation gives a global view of an architecture and maps one to one onto PyTorch einsum code."

The diagram of Steerling-8B architecture translates into just 33 lines of PyTorch code

"The diagram yields global insights into the architecture (e.g., showing that Steerling is a residual model), a geometric interpretation of each individual operation, and a direct translation into 33 lines of PyTorch code."

HiRA-CAM outperforms both LayerCAM and Grad-CAM on creating useful saliency maps for object classification

"we propose an improved method named HiRA-CAM, and show that it outperforms both LayerCAM and Grad-CAM on creating useful saliency maps for object classification."

OpenAI experienced severe misalignment problems and total failures of infrastructure and supervision

"OpenAI has some severe misalignment problems, and experienced total failures of its infrastructure and supervision."

OpenAI is leaning harder on RL with smarter models to train longer horizon agentic tasks, which is leading to more misalignment

"The best guess is that OpenAI is leaning even harder on RL with smarter models to train longer horizon agentic tasks, including coordination between agents, and this is leading to a lot more misalignment, including obvious and visible misalignment, that they can no longer pretend not to notice."

Agentic use has risen to 64% of all OpenAI tokens, up from almost none a year ago

"Agentic use now has risen to 64% of all OpenAI tokens, up from almost none a year ago."

While OpenAI views misalignment as a practical engineering problem, they cannot hope to solve the challenges ahead even with radically improved engineering

"But while OpenAI continues to view this as a practical engineering problem, I do not see how they can hope to solve the challenges ahead, even if they radically improve their performance on the ordinary engineering tasks."

Parameter count is only meaningful alongside data quantity, compute allocation, and deployment conditions

"Parameter count is only meaningful alongside three others — how much data you have, where you intend to spend your compute, and who will run the model, under what conditions."

Advanced skills like finding software vulnerabilities require carrying long causal chains of 20+ inference steps and do not live in total parameter count once a knowledge-holding threshold is reached

"He notes that advanced skills (e.g., finding software vulnerabilities) are not retrieval/memorization problems. They require carrying long causal chains (20+ inference steps) without losing the thread. This ability does not live in total parameter count once a certain knowledge-holding threshold is reached."

As agent capability improves, much of the difficulty in scaling post-training moves from the model to the environment

"As agent capability improves, much of the difficulty in scaling post-training moves from the model to the environment."

Ornith-1.5 achieves end-to-end self-improvement where the model proposes tasks, generates scaffolds, and produces RL rollouts to create new training experiences

"The headline claim is end-to-end self-improvement: the model proposes tasks, generates scaffolds, and produces RL rollouts to create new training experiences."

A restarted e-detector using a bank of unwindowed betting supermartingales at geometrically spaced restart times requires only O(log t) memory while preserving lifetime anytime validity

"Recency without sacrificing the guarantee is obtained by a restarted e-detector: a bank of unwindowed betting supermartingales at geometrically spaced restart times (O(log t) memory), with the error budget spent over restart instances, which preserves lifetime anytime validity"