I predict: Their AI is text-to-speech on images of AI-generated characters, animated in video, with lip-sync pasted from pre-recorded responses, and they will pretend it's live.
@dfintelligence
-
Debate on the strategic plan for AI in France at Delta Festival
By
–
I will debate the strategic plan for AI in France against @CBeaune (High Commissioner for Strategy and Planning) tomorrow at the Delta Festival in Marseille. What are the questions, comments, concerns or criticisms that strike you about the way AI is managed in
-
GPT5 preferred over Claude Code for complex tasks
By
–
Honestly I don't know for now. I use GPT5 much more than Claude Code for a week but it's because I make it work on ultra complex stuff like (create me a lite program to display networks in TypeScript in a fully optimized way). But after
-
Late-night coder shares 2-week AI-assisted setup: GPT-5 Pro & Claude Opus 4.1
By
–
It's 2:30 a.m., I'm coding, no one will read this tweet, but here's my computer setup for the past 2 weeks. I've got Codex with GPT-5 Pro running for ultra-complex tasks and one-shotting features for me (but in a really ugly way), and I've got Claude Opus 4.1 handling the
-
Managing AI teams for client projects, the freelancer’s dream
By
–
Very interesting, yes. I really think there is a world where we could all become managers of AI teams that code on independent client projects. The ultimate freelancer's dream. Honestly though, I don't think that will lower the price that much.
-
Speed Level Compared to Claude Code and Time Perception
By
–
Level speed isn't as long as Claude Code, I find. But the thing is, it talks A LOT. So I read A LOT. I don't know if it unconsciously makes me wait and gives me the impression that it goes faster lol.
-
GPT5 surpasses Claude Code in planning and complexity management
By
–
More info on why I say this: Task planning and complexity management ++ really doesn't scare GPT5 in pro mode. It is able to break down a very complex problem and find a very optimized solution. Whereas Claude Code for example it will
-
User: GPT5 with high reasoning surpasses Claude Code
By
–
Damn, I never thought I'd say this, but I think OpenAI's Codex with GPT5 on high reasoning is starting to be better than Claude Code.
-
Avoid frameworks in generative AI workflows
By
–
Il y a 2 ans, certaines équipes data ne juraient que par LangChain. Aujourd’hui, beaucoup s’en mordent les doigts. En IA, et notamment en IA Gen : N’UTILISEZ PAS DE FRAMEWORK. Et ce, pour une raison simple : les technologies, les outils, les providers, les méthodes… tout ça
-
The main problem: generative AI reduced to prompts and PDFs
By
–
Ah yes, and obviously the main problem: there must be a large portion of these companies that think doing generative AI means training on prompts and having a PDF of prompts.