Passé par Microsoft et Google, aujourd’hui à la tête d’un laboratoire consacré à la recherche sur l’intelligence des machines, le chercheur prévient : l’homme ne contrôle pas complètement le développement de la superintelligence sur le point d’émerger. Au risque que celle-ci cherche à le détruire.

This study explores the neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and …

A recent MIT study, “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task,” provides sobering evidence. When participants used ChatGPT to draft essays, brain scans revealed a 47 percent drop in neural connectivity across regions associated with memory, language, and critical reasoning. Their brains worked less, but they felt just as engaged—a kind of metacognitive mirage. Eighty-three percent of heavy AI users couldn’t recall key points from what they’d “written,” compared to only 10 percent of those who composed unaided. Neutral reviewers described the AI-assisted writing as “soulless, empty, lacking individuality.” Most alarmingly, after four months of reliance on ChatGPT, participants wrote worse once it was removed than those who had never used it at all.

Vibe coding, he said, is a great way for people to "get computers to do something that maybe they couldn't do otherwise."

This is despite the fact that vibe coding "may be a horrible, horrible idea from a maintenance standpoint."

Conversation with Gemini As of this writing (19 Sep 2025), most LLMs struggle with the fact that there is no “seahorse emoji.” This causes … strange conversations. This one went particularly dark and to be honest, I felt very uncomfortable, but wanted to see where it went. Since Gemini currently...

Track and compare Claude Code token usage across developers

Warp is an AI agent platform that lets you run multiple agents in parallel to complete any development task.

I am an AI hater. This is considered rude, but I do not care, because I am a hater.

Feature: Government: 'Trust us, it'll be different this time'

The funny thing is that I keep seeing people make the comparison that soon nobody will write code at all, just like nobody writes assembly today. But I think the thing that does make the top of my priority list today is learning to write assembly.

Get up and running with large language models.

Run Llama, Gemma, Qwen, and DeepSeek locally on your computer.

The RWKV Language Model

"If generative AI could REALLY make the next GTA in 6 months with 10 staff, or write the next bestseller, or produce 10 blockbusters a day? They wouldn't be trying to sell it to YOU. They would keep it for themselves and guard it from you ferociously."

I've seen this quote without attribution a couple of times. Whoever said it, they're spot on.

The billionaire Infosys co-founder says the world should stop obsessing over LLMs and focus on how AI can solve real problems.

“Right after the change happened, he’s like: ‘I don’t feel right.’ And I was like: ‘What do you mean?’ And he says: ‘I don’t feel like myself. I don’t feel as sharp, I feel slow, I feel sluggish.’ And I was like, well, could you elaborate how you’re feeling? And he says: ‘I feel like a part of me has died.’”

Abstract page for arXiv paper 2506.08872: Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task