Holy shit… Meta just dropped a paper that flips the “AI will improve itself and leave us behind” narrative on its head and the implications are massive Here’s the wild part: They argue the safest and fastest path to superintelligence isn’t self-improving AI at all. It’s
ETHICS
-
AI Comparison: The Psychopath Method
By
–
I saw a guy prompting today.
Tab 1 Claude.
Tab 2 Gemini.
Tab 3 ChatGPT.
Tab 4 Grok.
Tab 5 DeepSeek.
He asked every AI the same exact question.
Hit run on all five.
Patiently waited, then compared each model’s response. Picked the best one.
Like a psychopath. It's me. -
Wadhwani AI’s Last-Mile Solutions Transform India’s Agriculture Education Healthcare
By
–
India's growth story is supported by Wadhwani AI’s last-mile solutions for agriculture, education and healthcare. We collaborate with frontline workers, educators and communities because teamwork makes the dream work! #WadhwaniAI #AIforBharat #AIforGood #EdTech
-
Engineering Emergent Personalities in LLM Reward Optimization
By
–
There is definitely work going into engineering the "you" simulation – the personality that gets all the rewards in verifiable problems, or all the upvotes from users/judge LLMs, or mimics the responses of SFT, and there is an emergent composite personality from that. My point is
-
Should France ban certain speech online?
By
–
Vous pensez que ce discours doit être interdit en France ?
— Docteur Laurent Alexandre (@dr_l_alexandre) 7 décembre 2025
pic.twitter.com/Hgma8lsPywVous pensez que ce discours doit être interdit en France ?
-
Study on Human-AI Synergy in 2025
By
–
Paper: "Quantifying Human-AI Synergy" by Riedl & Weidmann, 2025 https://openreview.net/pdf/8b876352f6a3393c76a956a97f940c0512e38671.pdf
… -
Collaboration, Not AI, Is the Real Bottleneck
By
–
here's what broke my brain about this research. we've been optimizing the wrong side of the equation this entire time. better prompts. stronger models. higher benchmarks. longer context windows. more parameters. but the bottleneck isn't the AI. it's our ability to collaborate
-
Not Everything Is an Engineering Problem in AI
By
–
Thus not everything is an engineering problem, at least for us.
-
Learning attendance as a solution to a serious issue
By
–
Thanks Josh. If it truly had learned how to attend this wouldn’t be as serious an issue?
-
Claude’s Post-Training Excellence and Alignment in AI Models
By
–
Every time you are chatting with claude(Opus 4.5 esp. but seems to apply to all models), you quickly realize how clever claude post-training is. The attention to details, data quality. It's rare that claude will break in the middle of conversations. I think this is beyond system prompt and training dynamics/physics, and more about aura-like things: behaviors, personalities, when to push back/when to defer, and countless subtle edge cases. It's no surprise alignment people there follow closely the training. There was an invited talk at CMU this past spring on alignment and I asked why claude vibes and post-training feel different, the answer was high-level as you'd expect, but same: training, post-training, alignment, and evals teams are closed-loop. Sam Bowman (@sleepinyourhat) From everything we know so far, Opus 4.5 seems to be the best-aligned model out there in a bunch of ways. I follow the training process closely as part of my work on alignment evaluations. Here's my guess about the two things that are most responsible for making 4.5 special. 🧵 — https://nitter.net/sleepinyourhat/status/1997006353647522098#m