🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

"Is it possible to articulate a conception of consciousness that is compatible with the exotic characteristics of contemporary, disembodied AI systems, and that can stand up to philosophical scrutiny?" Updated version of this paper on arXiv.org: arxiv.org/pdf/2503.16348

great to chat w Tyler for this list!

Gemini 2.5 Pro shipping a granular thinking budget that works could actually be a pretty big deal. The glorious slider that comes before the model just knows how much thinking to do. Helps limit overthinking, helps collect good data on "how hard users think the model is"

blog in 2-3days

People in the mega online AI community here take it for granted how important this is and how few people actually know the scale of adoption that AI has. One of the biggest stories of the year.

Loved this point from the new Interconnects "Not all tokens are created equal, but this is about 150-200M tokens per second. In a world with 5T Google searches annually, which translates to around 100K searches/second, that tokens per second number is equivalent to roughly

Mabe the hyperstition is working j⧉nus

Nudging vs Boosting Listen to my colleague Ralph Hertwig's latest. youtube.com/watch?v=z9r-TJ…

Is this real?

Foresight has just launched AI Safety RFPs. They fund $5M in grants annually, and are now seeking projects across 39 areas within neuro x AI, AI security, AI-enabled coordination, research & forecasting automation. Next application deadline is June 30th. foresight.org/request-for-pr…

High capital costs (e.g., in filmmaking) push products towards popular culture to recoup investments, while low capital costs (e.g., in painting) allow for more avant-garde and niche pursuits. If AI massively decreases production costs and other barriers to entry, I would expect

immortalizing this moment forever when RL is so easy that you can just use random rewards and your benchmarks still go up smh

Every time I've underestimated the symbolic weight of something has been at my own peril

As we do societal evals at CIP —public health, AI relationships, democracy, etc. across regional languages we've spent a lot of time dealing with how brittle LLM judge pipelines are. Stoked to share an open-source test suite (blog + code) we’ve built to stress-test ours before

🤷‍♂️

Another great paper from Hannah and the team. The part on social reward hacking is especially interesting!

Lack of RL Logical Reasoning data? Excited to share SynLogic: Synthesizing Verifiable Reasoning Data at Scale for Learning Logical Reasoning and Beyond 🚀 Building strong logical reasoning through RLVR 📄Paper: huggingface.co/papers/2505.19… 💻 Code: github.com/MiniMax-AI/Syn… (1/n)

Our interpretability team recently released research that traced the thoughts of a large language model. Now we’re open-sourcing the method. Researchers can generate “attribution graphs” like those in our study, and explore them interactively.

Google DeepMind just put out a Lean 4 repo of formalized open math problems. We need more contributions to the database! github.com/google-deepmin…