The robots.txt will be able to define __usage_ai-training to target AI crawlers.

I will also make a prediction: as AI will fall short or costs way to much, the documentation will have to improve drastically. The standard will become Astro or Stripe at the minimum.

Comment assumer ensuite le projet en production alors qu'il a été vibe-codé?

Et l'effet se cumule. Une application de plus, c'est un dépôt, un environnement, des dépendances, des accès et un cycle de vie supplémentaires. À force, nous constituons un patrimoine de code dont la maintenance dépasse les moyens disponibles.

Another techbro take that is dangerous for people. They created the shortage in RAM.

I didn't have use cases for this, but it's interesting to see how massive text data are processed. The goal was to build a tool FineWeb with high quality data.

L'IA ayant aspiré tout Internet peut faire des croisements inédits. Ici pour retrouver un film à partir d'un prompt:

je recherche un film des années 80. Voici les éléments : c’est dans un monde post apocalyptique. Un homme est crucifié dans le désert. Sur le sol. Il se bande ensuite les mains. Il y a une scène où il se fabrique un nunchaku et il l’utilise ensuite dans la seconde partie du film. Quel est ce film ?

It makes sense to believe that all the tricks we come up with with to make models behave better, the ways of working with models that get us to a useful result faster, if they work, are going to be folded back into the product, packaged, and become the default experience a few months later.

Maybe I'm getting old, but in a world where Google, Facebook, the iPod, iPhone and iPad, and so many other examples have shown that getting it right is more valuable than getting in first, I feel content to just keep pace with the pack

La réponse de la société mathématiques de France à la publication de plus de sept cents textes mathématiques produits par ChatGPT.

My feeling is that, overall, the situation here is a bit analogous to the discussion here about why I don't like when people talk about how AI gives a 100x or 1000x productivity increase. There are some things that are sped up by a very large factor, easily over 1000000x. And some things don't currently get sped up at all.

This would be the best occasion to release all images to the creative common licences

Getty Images – which has supplied photos and video for news agencies for more than 30 years – has been struggling amid a pile of debt and mounting competition from AI image generators.

The bankruptcy occurs because AI generated images are used instead of relying on suppliers.

The principal resource is capital and only capital.

Linus is saying the quiet part out loud: AI-native projects are environments where the key ingredient is capital. Abstraction can be performed by a quiet appreciation. True automation can be disruptive.

AI-based architectures (again, assuming that the AI works reliably, which is a highly contingent proposition) translate all that stuff into capital flows. A "machine speed" project requires money to participate in, regardless of their complexity.

AI usage starts to cost. 29$ per employee for Accenture.

Christian was part of a team behind a landmark study that found that spending just 10 minutes relying on an AI tool significantly impairs our ability to focus and persist with difficult tasks

Additionally the "fix" fixes more than 40 separate issues in a single mega-commit touching 68 files which like. This is really beyond "you are trusting the slop machine too much". This is "the moment the slop machine got in your hands, a switch flipped in your head and you were like 'I never have to follow any engineering standards, any at all, ever again'". To repeat once again: "AI" is not a technology, it is a permission structure.

The good engineering practices are ignored by AI and the people seems to not care about it.

Des réflexions sur la stabilité de quelque chose développé par IA. L'exemple est celui de ArtCraft: https://shaarli.lyokolux.space/shaare/aaHWxg)