• Generative AI has polluted the data
• Information that used to be free became expensive

If someone is collecting all the text from your books, articles, Web site, or public posts, it's very likely because they are creating a plagiarism machine that will claim your words as its own.

LinkedIn content is now used to train AI.

There’s some good news for users in the EU, the UK, Iceland, Norway, Liechtenstein (both of them!) and Switzerland as their data isn’t being used to train LinkedIn's AI at all and won't for the foreseeable future.

The downside of AI for programming:

• Erosion of Core Programming Skills
• Over-Reliance on Auto-Generated Code
• Lack of Ownership and Responsibility
• Reduced Learning Opportunities
• Narrowed Creative Thinking
• Dependency on Proprietary Tools
• false sense of expertise

Context:

To understand what’s happening here you need to remember that it’s a category error to treat LLMs as thinking entities.
They are statistical models that work with numbers – tokens – that represent language and the relationships between the words. It’s statistics about language wrapped up in an anthropomorphic simulation.

Attack:

The token stream (Strategic Text Sequence) itself – the numbers not the words – is an attack surface.

Reality of the threat:

This is going to get automated, weaponised, and industrialised. Tech companies have placed chatbots at the centre of our information ecosystems and butchered their products to push them front and centre. The incentives for bad actors to try to game them are enormous and they are capable of making incredibly sophisticated tools for their purposes.

I’ll point out that the training data requires the wholesale harvesting of creative works without compensation.

I’ll also point out the ludicrously profligate energy use required not just for the training, but for the subsequent queries.

but "these things will get better!"... first there is no evidence. Second what the hell kind of logic is that?

Tout écrivain, Saint-Exupéry le premier, vous le dira : l’art de l’écriture, c’est de supprimer, de trancher, de raccourcir le texte pour lui donner de la puissance.

Dans mon entourage, les gens l’utilisent pour envoyer des dossiers administratifs. Alors, est-ce utile ? Non, c’est juste que ces dossiers sont complètement cons, que personne ne va les lire et qu’on utilise des outils cons pour gérer des problèmes à la con qu’on se crée soi-même.

Un outil et non une solution:

Comme le dit la linguiste Emily M. Bender, on ne demande pas aux étudiants de faire des rédactions parce que le monde a besoin de rédactions. Mais pour apprendre aux élèves à structurer leurs pensées, à être critiques. Utiliser ChatGPT c’est, selon les mots de Ted Chiang, prendre un chariot élévateur à la salle de musculation. Oui, les poids vont faire des va-et-vient, mais va-t-on se muscler pour autant ?

aphorisme: « tout le monde peut écrire, l’écrivain est celui qui ne sait pas s’empêcher d’écrire »

Bruno Leyval dessine tous les jours depuis qu’il est tout petit. Il dessine tout le temps. Il s’est transformé en machine à dessiner. Cette sensibilité de toute une vie ne pourra jamais se comparer à un algorithme générateur d’images.

À propos de l'IA qui génère du code: On cherche à optimiser la « création de logiciel » tout en oubliant la maintenance du logiciel et de l’infrastructure pour le faire tourner.

It's perfectly legal for people to take your writing, code, videos, music and other works into a 'dataset' that can be used to train an LLM model to forge your art or writing style -- for money.

But if a nonprofit decides to purchase hardcopy books, scan them in, and create a digital lending program providing works to anyone who asks -- for free, that's checks notes illegal. :D

If it's ok for OpenAI, then it is ok for the rest. Maybe it is not.

The question is: why should I buy something while OpenAI use it for free? Where do you draw the lines? Is piracy now fair? What's the difference between OpenAI and piracy?

CoPilot can be encouraged to launch HTTP requests on the server side, potentially enabling access to data from other companies.

A server-side request forgery (SSRF) bug in Microsoft's tool for creating custom AI chatbots potentially exposed info across multiple tenants within cloud environments.

Though the research proved inconclusive about the extent that the flaw could be exploited to gain access to sensitive cloud data

Microsoft y indique qu'ils mettent simplement un outil à disposition et que l'utilisateur est responsable de tout: droit d'auteur, .

Microsoft stocke des informations sur sa création et associe ces informations et le contenu à un identifiant de contenu

Article anglais original de la dépêche de NextImpact: https://www.theregister.com/2024/08/14/microsoft_services_agreement_update_warns/

Thoughts on AI usage: do they have a positive or negative global impact?

“Jim Covello, Goldman Sachs’s head of global equity research, told me, “If we’re going to justify a trillion or more dollars of investment, AI needs to solve complex problems and enable us to do things we haven’t been able to do before.”

Yes it's a quote of a quote

O_O it performs better than xz