It's a stance; never about quality, voice and/or style, around for other human made products.

I 100% agree.

It's useless to argue with "by AI". What is it solving for me? Is it good and reliable for me?

And it doesn’t really matter if these features are branded as “AI”, “smart” or “automation”. However, they must work well for people using them. And that means that people must be aware of use cases where it actually helps them, and be inspired to find more use cases on their own.

Sans surprise

We don't have to follow the trends.

I've started asking clients a simple question when they bring it up. Not to be difficult, just to understand: "Do you actually use chatbots when you visit other websites?"

Simple is the word that keeps coming up. And I've learned that when a client says simple, they don't mean easy to use. They mean not impressive enough.

The pressure isn't really coming from clients anyway. It's coming from the web itself, from a decade of bloated pages, dark patterns, and feature arms races that quietly redefined what a "real" website looks like. Clients are just reading the room. The room is wrong, but they're not imagining it.
The shift might come from users, not decision-makers. It might come when enough people notice that the fast, calm site was easier to use. That they actually found what they came for. That they didn't have to close three things before reading a single line.

With some 300 data centers already in operation and 200 more in development,

80 GW of new gas plants into construction. The half is reserved for data centers.

The colossal Stargate project is using minor permits, that allows 62 backup diesel generators. That's 1.6 million tons of CO2a and 1000 tons of air pollutants. It's not minor anymore.

In several cases, data centers secured these permits before seeking massive expansions later on, deploying a “small first, big later” strategy that watchdogs say limits public input and creates unstoppable momentum for their projects.

Stargate expanded for example the next year 4 times more turbines and 30% increase of generators.

[W]e will become deeply skeptical of one another’s realness. Every time you find a new favourite blog or Twitter account or TikTok personality online, you’ll have to ask: Is this really a whole human with a rich and complex life like mine? Is there a being on the other end of this web interface I can form a relationship with?

Ms. Appleton explains how we can use the limitations of language models to prove our humanity:

This leaves us with some low-hanging fruit for humanness. We can tell richly detailed stories grounded in our specific contexts and cultures: place names, sensual descriptions, local knowledge, and, well the je ne sais quoi of being alive. Language models can decently mimic this style of writing but most don’t without extensive prompt engineering. They stick to generics. They hedge. They leave out details. They have trouble maintaining a coherent sense of self over thousands of words.

Also human beings can vouch for other human beings using classic web methods such as blogrolls or webrings, or by linking to other interesting articles.

Un retour sur des outils d'IA pour la sécurité.

I love the notion of “Quiet AI”. These are tools that are mostly invisible, sit in the background, and do small tasks on the user’s behalf. They never scream for attention but happily assist in repetitive, frustrating tasks that can easily be automated or assisted with a smart helper.

Yes because AI is always a tool and not a solution by itself. What should I do as user with a "Magic AI buttons"; rewrite the button as "Add automatic transcription" (with AI powering the feature) and everyone will understand.

The take here is to document how each folder should behave. An AI agent can take care of the rest. Its integration is seamless.

Well unless the AI agent can not reproduce the same behaviors but it's another discussion.

For transparency

How Google is slowly killing the web and

If you do care about the web, about people’s ability to participate in it as more than mere passive consumers, this needs to be taken seriously. De-googlifying your mental apparatus becomes more urgent today. Find other search engines, don’t use the Chrome browser. Or wake up in a slopified AOL kind of environment where your access to information is limited to what Google’s synthetic text extruders deem relevant.

Degooleify

When ChatGPT appeared in 2022, I quickly realized the Web was being forced into another battle. This one is different. With BBSs, the Web won because it was more accessible. With Flash, the Web won because open standards eventually became powerful enough. With mobile apps, the Web survived because search, links, and publishing were still essential.

This is the browser losing its position as the primary interface to knowledge.

Stack Overflow was the warning shot of the web searches being replaced by AI.

The website will becomes even more infrastructure: Websites will increasingly become infrastructure for machines. They will feed models, agents, search systems, APIs, datasets, crawlers, and private knowledge bases. A source of truth and data.

Emails will remain, but younger generations do not think in folders, inboxes, subjects and signatures. They think in threads, voice notes, reactions and instant replies.

If people stop browsing, stop searching, and stop clicking, then visibility depends on whether AI systems decide your content matters.

Then browsing and URLs will be lost little by little.

People will not say, "Today I reject the open Web." They will simply ask the assistant because it is faster. Convenience wins. Convenience replaces nostalgy, all the time. It almost always wins. The Web's biggest enemy is not ideology. It is not regulation (well, a bit yes). It is not even AI itself. It is convenience.

So the web can become a nerd medium again, like IRC, FTP, Gopher or Bulletin Board System.

Now I am watching AI become the next interface. The pattern is obvious. Interfaces change. Behaviors remain. People want answers. People want connection. People want tools. People want convenience. People want to publish, learn, buy, flirt, argue, create, complain, and feel less alone with their questions. The container changes. The hunger does not. The Web we know is going to disappear. Not because it failed, but because it succeeded so completely that its content can now be absorbed into the next interface.

Maybe this article will be read by humans. Maybe it will be summarized by an AI into two polite lines for someone curious on the other side of the planet. Maybe that person will never visit this page. That is probably the future. A little sad. A little funny. Very predictable.

Use UIs instead of writing code. Ok.

I don't believe this is the case. In my opinion, LLMs do not represent a layer of abstraction, but rather a layer of automation. It doesn't abstract away the code layer, it automates it.

I believe in the importance of making understandable software. To this end, I'm working on projects that further this goal.

Primarily, I develop ReTangled, an Entangled-compatible bi-directional tangler written in Rust. It's meant to extend literate programming as far as it can go while integrating it as tightly into existing toolchains as reasonable. As of writing, it's in the early stages of development. I welcome contributions!

https://codeberg.org/liberty/retangled

Des palettes de vieux livres venant de librairies du monde entier sont expédiées vers les Etats-Unis. Des entreprises spécialisées en IA seraient à l'origine de cet accaparement massif de patrimoine. Elles utiliseraient ces livres comme données brutes pour entraîner leurs modèles de langage, puis les jetteraient après leur numérisation.

Le problème est bien de les jeter ensuite. Le mininum serait de les rediffuser en copie numérique, ou de les renvoyer.

L'entreprise de ce business serait "Zoom Books": Nous ciblons des ouvrages de non-fiction publiés à partir de 1970 et dotés d'un numéro ISBN – des invendus poussiéreux dont personne ne voulait depuis des années."

Quiconque numérise des textes et les diffuse sur Internet s'expose à des poursuites pour violation du droit d'auteur. Mais, aux Etats-Unis, la législation autorise l'entraînement de modèles d'IA à l'aide de livres acquis légalement. Si une entreprise achète de grandes quantités de livres d'occasion à bas prix, les désassemble et les numérise, les textes peuvent ensuite servir à entraîner des modèles d'IA.
C'est le principe étatsunien du "fair use", ou principe d'utilisation équitable. Il autorise l'utilisation d'œuvres protégées par le droit d'auteur sans l'autorisation expresse du titulaire des droits, à condition que cette utilisation serve l'éducation du public et stimule la production intellectuelle.

Parce que oui

À court terme, les libraires se réjouissent d'écouler leurs invendus. À long terme, cependant, un tout autre scénario se dessine. Si ces anciens livres sont détruits à grande échelle, ce patrimoine sera irrémédiablement concentré sous la forme d'une masse de données numériques possédée par quelques entreprises. Un accaparement discret, systématique et soumis à aucune délibération publique.

Et le copyright ne s'appliquera plus sur les livres ingurgités par les IA.

A layer against AI to make copyright right.

More about it at https://www.wired.com/story/matthew-butterick-ai-copyright-lawsuits-openai-meta/

Donc l'État a arrrêté après 6 mois un contrat sur 3 ans. Bravo la stratégie.

Aucune précision n’a en revanche été donnée sur les modalités concrètes de cette passation au sein de la DGSI, alors que le contrat avec Palantir courait théoriquement jusqu’en 2028.

Because it has other purposes

Ähnlich zu https://shaarli.lyokolux.space/shaare/vP775Q sind KI-Modellen nicht deterministisch. Das ist halt ein Problem.

Das Unterhnemen erklärt seinen Punkt: kein Agentic Engineering ist geplant.