Loading activity…
@allenainie
Research Scientist @GoogleDeepMind. Agentic post-training and CUA. Co-creator of Trace. Prev: RL PhD @StanfordAILab, @MSFTResearch, @DeepMind, @AWS Neuron
AI
Arora shared defense slides titled “Causal Interpretability for Analysing and Improving Language Model Architectures.”
AI
A prelaunch tester praises Gemini 4's creativity and writing but says it lags at complex multi-agent work.
AI
A user expects releases from Google, Anthropic and OpenAI that week, with xAI and Moonshot news also possible.
AI
Arthur Douillard describes work since March to scale reinforcement learning for Gemini, with bug fixes, improvements and more still to build.
AI
Nado’s Sept. 30 quote post pairs an “it’s so over” / “we’re so back” joke with Hayden Evan’s older criticism of 3.7 Flash. Andrew Curran adds “Status: back.”
AI
Google announces a 1 million-token output limit, up from 64,000, for its new model. Initial access goes to trusted cyber defenders through the Fairwind Program.
AI
Sundar Pichai says AlphaGenome Atlas mapped all 9 billion possible single-letter genetic changes across the human genome. His update also highlights an open-access report on global AI use.
AI
Anthropic says its model produced the first end-to-end computer-checked version of the theorem.
AI
New Flash models target agentic workflows and cybersecurity tasks with API and enterprise access.
AI
Former Google Brain resident focuses on reinforcement learning and post-training at DeepMind.
AI
Google's new model improves coding accuracy and agentic tasks at reduced cost.
AI
Four former Google researchers found a public benefit corporation to automate machine learning and science.
AI
Recent MIT PhD graduate moves to San Francisco for role focusing on pretraining and AI societal impacts.
