I'm excited to share MindScript, a new programming language that gives both formal and informal computation first-class status, blending them smoothly. Check out the Github repository for more details:

Great interview of the creator of Redis (Salvatore Sanfilippo). A true hacker who gave a massive present to the world, and cloud providers make billions of it. The irony. (Sadly the interview is in Italian).

Generate videos in just a few seconds. Try Grok Imagine, free for a limited time.

RT @METR_Evals: We ran a randomized controlled trial to see how much AI coding tools speed up experienced open-source developers. The resu….

RT @pfau: Proud to announce that my stealth team has made a breakthrough that will unlock the next leap: artificial HYPER intelligence, whi….

The Messiah Complex is so last millennium. Welcome the John Connor Complex, where you’re convinced you’ve been sent back in time to lead the Resistance, hack a T-800 with a MacBook, and breakup the rise of killer robots.

RT @getjonwithit: Calling c the "speed of light" completely misses the point. Rather, c is the "spacetime exchange rate": how many units of….

I am but a tiny red robot.

RT @pfau: More high weirdness in training models to reason with RL.

RT @vasumanmoza: Claude 4 just refactored my entire codebase in one call. 25 tool invocations. 3,000+ new lines. 12 brand new files. It m….

RT @JQ_Zhu: 1/11 Controlling LLM behavior in risk-sensitive, high-stakes domains like healthcare & finance is crucial for safety. In our ne….

RT @daiosai: Don’t be fooled by Big Tech’s “open protocols” (e.g. MCP) – they’re marketing hooks built to lock you in. Stand up for the We….

Paper:

. please present rigorous arguments that demonstrate their limitations. Too many arguments rely on informal philosophical critique or broad appeals to psychology, machine learning, or information theory, and don't even engage with the LH definition.

There is only one well-established formal measure of intelligence —the Legg–Hutter definition— grounded in Kolmogorov complexity and subjective expected utility. If you question either of these foundations, .

RT @g_k_swamy: 1.5 yrs ago, we set out to answer a seemingly simple question: what are we *actually* getting out of RL in fine-tuning? I'm….

RT @driainmurray: Image models struggle to create Go boards. Sadly ChatGPT can't currently copy one either, it seems conditioning the mode….

RT @pfau: New paper accepted to ICML! We present a novel policy optimization algorithm for continuous control with a simple closed form whi….

RT @DimitrisPapail: We’ve been cooking. a new open weights 14B Phi-4 reasoning model, SFT’d on ~1.4M carefully curated reasoning demonstr….

Excellent initiative!.

Whenever I read a "reward is enough" position paper I ask myself whether we're still in 1985 in ML-terms or worse, 1954 in economic theory terms.

RT @danijarh: Excited to share that DreamerV3 has been published in Nature!. Dreamer solves control tasks by imagining the future outcomes….