Topic

#Ai Research

11 articles on Ai Research — news, releases, guides and analysis from the SourceFeed engine.

When the Agent Outgrows Its Judge, Evolve the Judge

Cambridge and NVIDIA's Red Queen Godel Machine makes the evaluator part of the self-improvement loop, with guardrails worth stealing.

A 5B model raised strictly on grade-school text shrugs off every trick meant to push it further.

A model raised on nothing past fifth grade shows RL, scaling, and prompting only amplify what pretraining put there.

OpenAI's first telemetry-based study shows broad but shallow adoption, driven by juniors and rich firms.

An unreleased model jumped a 50-year-old bound from 41.6% to 67.2% inside Claude Code, with a machine-checked proof.

Alphabet is funding its legends on their way out and handing Gemini to a product-first org.

Terence Tao's ChatGPT transcript shows why the same model does frontier math for experts and boilerplate for everyone else.

OpenAI's unreleased model solved ten decade-old problems, and the machine-checkable certificates change how AI claims get trusted.

$OpenAI Disproved an Erdős Conjecture, and the Math Checks Out$

Named mathematicians verified the counterexample — the first credible autonomous AI research result, seven months after a false alarm.

A Stanford-led analysis puts hard numbers on the frontier's retreat from publishing, and on who filled the gap.

The recovered HAWK-256 key is a demo; the reduction behind it pushes HAWK-512 under NIST's Category 1 bar.