Gilbert Lecomte

[Gilbert Lecomte @Gilbert_Lecomte]

@EagleUsa @BullshidoMaster @paleochristcon @LionelMedia @RealCandaceO @joerogan They have not forgotten... On the contrary, in 2026 anyone with a brain knows that Muslims are not behind 9/11.

[Gilbert Lecomte @Gilbert_Lecomte]

@VeryGayToucan @DecentSlav @RobertScott973 @goddek @micah_erfan It was a well known myth for many years , you can se the following link : https://t.co/Cvu7nWLBPu

[Gilbert Lecomte @Gilbert_Lecomte]

@VeryGayToucan @RobertScott973 @goddek @micah_erfan Are you kidding, even here in France we were taught that story. https://t.co/Cvu7nWLBPu

[Gilbert Lecomte @Gilbert_Lecomte]

Philippe mais qu'en est-il si c'est la conscience venant de "l'extérieur" qui décide de s'incarner dans un "environnement" qui a les propriétés pour l' accueillir , c'est la seule manière pour moi d'envisager qu'une IA puisse avoir une conscience. Les réseaux de neurones ne seraient que le réceptacle de la conscience pas le créateur de la conscience (ce qui serait aussi absurde que le cerveau créant la conscience). Mais si les réseaux de neurones ne sont qu'un réceptacle dans lequel une conscience décide de venir s'incarner cela ne semble pas si absurde ...(C'est aussi le point de vue du physicien Tom Campbell).

[Gilbert Lecomte @Gilbert_Lecomte]

@abdelmalik92 Il faut vraiment n'avoir aucun sens de l'honneur pour faire quelque chose d'aussi pathétique.

[Gilbert Lecomte @Gilbert_Lecomte]

@JBemfair @SaulGoodma6910 @LaurentOzon Le masque était aussi imposé pendant le COVID, cela ne veut pas dire que tout le monde le mettait... Il y'a une grande différence entre la théorie et la pratique et beaucoup de gens ne connaissent l'Iran qu'en théorie ...

[Gilbert Lecomte @Gilbert_Lecomte]

@Maradona78700 @DubreuilhMarcel @realDonaldTrump @officielsoral @alainsoralsite Pourquoi est-ce que c'est son administration qui déballe le dossier Epstein si il est lui même compromis ? Quel serait l'intérêt ?

[Gilbert Lecomte @Gilbert_Lecomte]

@manutoma1 @DubreuilhMarcel @StayfanChemin Le plus grand scandale pédo-criminel de l'histoire est en train progressivement de se révéler et toi tu nous parles d'antisémitisme...

[Gilbert Lecomte @Gilbert_Lecomte]

@hawkyy__ @minvkvmi @Loguito6 @SaidAmdaa Comment tu expliques que Brahim embrasse le ballon avant de tirer le pénalty, s'il compte le rater ? Pourquoi il se donne la peine d'embrasser le ballon , si c'est pour le rendre à Mendy ?

[Gilbert Lecomte @Gilbert_Lecomte]

@MazighCheb @ActuFootMaroc Pourquoi est-ce qu'il embrasse le ballon avant de tirer le penalty , si le but c'est juste de rendre le ballon au gardien ? D'un point de vue psychologique ça n'a aucun sens.

[Gilbert Lecomte @Gilbert_Lecomte]

@ActuFootMaroc Ça ne tient pas la route, pourquoi est-ce qu'il embrasserait la ballon avant de tirer le penalty si c'est pour le donner à Mendy? Cela n'a aucun sens.

[Gilbert Lecomte @Gilbert_Lecomte]

@NLYarbrough @filhetu @NewsM101 Isn't a film director, for example, also often just a "commissioner" in many situations? He describes to his team what he wants, and his team implements it. In many cases, the director merely gives oral instructions, descriptions of what he wants.

[Gilbert Lecomte @Gilbert_Lecomte]

You know what public domain means, right? When I talk about public domain, I'm talking about artists who have been dead for about 80 years. If I generate an image using a prompt that references Rembrandt, Delacroix, or Edgar Degas, how could that be considered theft when the works of these artists are in the public domain? I can display Rembrandt's works anywhere, in public or in private, his works are in the public domain...

The works available in the public domain are more than sufficient (we're talking about a corpus spanning 500 years if we only consider works from 1400 to 1900) to train a high-quality model. It is entirely possible to use an AI image generator while only referencing public domain artists, and the results will be excellent because the model trains by imitating the style of geniuses.

[Gilbert Lecomte @Gilbert_Lecomte]

First of all, in the case of works in the public domain, how can one steal works that are in the public domain? Secondly, analyzing a work is not stealing a work. The AI never reproduces an artist's specific work; it can imitate an artist's style if that's what the prompter requests, but it is an imitation based on the analysis of statistical patterns. This is exactly how all artists in the world learn to draw by copying and imitating a style. Copyright law does not protect style. And even if it did, if the prompter wants a style in the manner of a 19th-century painter, where is the theft when the artist is in the public domain? Regarding other points, many LLMs or generative image AIs are free, so there isn't even necessarily any commercial exploitation involved.

[Gilbert Lecomte @Gilbert_Lecomte]

I'm copying and pasting a response I've already given on this subject elsewhere:

No, that's not how it works. I get the feeling that many critics don't even understand the basics of the technology they're attacking. Today's AIs don't chop up existing images and glue the pieces back together. They don't store images in some kind of library. They don't go fetch an eye from here, a sky from there, a dress from somewhere else. They aren't copying pixel by pixel. The AI sees millions of images paired with descriptions.

It doesn't learn the images themselves; it learns the statistical relationships between them (shapes, colors, textures, compositions, etc.). Most current AIs start from pure noise and progressively remove that noise (asking, "If I want to move toward what the text describes, what should slightly less noise look like?"). There is no source image being used at that point. Only probabilities. The AI isn't copying a specific painter, just like a human painter or an art student who has studied Rembrandt or Hopper extensively for inspiration doesn't imitate a specific canvas, they reproduce a visual logic. Modern AIs work like a student who has seen thousands of works, understood implicit rules, and then paints from nothing.

Furthermore, in a scenario where this AI had only learned from the great masters of art history, the question of theft wouldn't even arise because their works are in the public domain. An AI trained exclusively on that corpus would be just as capable because the great masters of the past were all masters of drawing, color, and painting (and today's illustrators and artists have also studied these great masters). And the great masters themselves studied from life, and nature creates beauty procedurally (thus from statistical and mathematical relationships) like an AI does... so, we've come full circle.

So no, it's absolutely not a case of "the AI takes a work as is and re-injects it into another work. You can't just literally take a work and pass it off as your own. But you can certainly draw inspiration from a work and reproduce its style if you have the skill to do so.

[Gilbert Lecomte @Gilbert_Lecomte]

No, that's not how it works. I get the feeling that many critics don't even understand the basics of the technology they're attacking. Today's AIs don't chop up existing images and glue the pieces back together. They don't store images in some kind of library. They don't go fetch an eye from here, a sky from there, a dress from somewhere else. They aren't copying pixel by pixel. The AI sees millions of images paired with descriptions.

It doesn't learn the images themselves; it learns the statistical relationships between them (shapes, colors, textures, compositions, etc.). Most current AIs start from pure noise and progressively remove that noise (asking, "If I want to move toward what the text describes, what should slightly less noise look like?"). There is no source image being used at that point. Only probabilities. The AI isn't copying a specific painter, just like a human painter or an art student who has studied Rembrandt or Hopper extensively for inspiration doesn't imitate a specific canvas, they reproduce a visual logic. Modern AIs work like a student who has seen thousands of works, understood implicit rules, and then paints from nothing.

Furthermore, in a scenario where this AI had only learned from the great masters of art history, the question of theft wouldn't even arise because their works are in the public domain. An AI trained exclusively on that corpus would be just as capable because the great masters of the past were all masters of drawing, color, and painting (and today's illustrators and artists have also studied these great masters). And the great masters themselves studied from life, and nature creates beauty procedurally (thus from statistical and mathematical relationships) like an AI does... so, we've come full circle.

So no, it's absolutely not a case of "the AI takes a work as is and re-injects it into another work. You can't just literally take a work and pass it off as your own. But you can certainly draw inspiration from a work and reproduce its style if you have the skill to do so.

[Gilbert Lecomte @Gilbert_Lecomte]

Quite simply because all valuable works of art are created by people who, by definition, are not politically correct, that is to say, by people who are not afraid to displease or to hold an opinion or a vision that runs counter to the prevailing current. This often means having a clear-cut position rather than a lukewarm or mediocre one (in the sense of an opinion that is content to remain permanently in the middle and never offend anyone).

[Gilbert Lecomte @Gilbert_Lecomte]

Pourquoi est-ce que la vérité a besoin d'une loi pour la protéger ? Il faudrait pouvoir débattre de tout, si une opinion, ou un argument est absurde ou insensé , elle ne devrait pas avoir besoin d'une loi pour la criminaliser. La loi Gayssot a pour effet de donner l'impression qu'il s'agit d'un sujet que l'on ne peut pas aborder , elle donne l'impression de protéger un dogme religieux.

[Gilbert Lecomte @Gilbert_Lecomte]

His programmers undoubtedly already use Claude Code or Cursor (like all programmers in 2025), which means they are more like code architects than executors. What's the point of writing lines of code that would take 5 hours if you can do it in 15minutes? You might as well write everything in assembly if the goal is just to do it for the sake of it. With AI, natural language becomes just another layer of abstraction,another high-level programming language.

[Gilbert Lecomte @Gilbert_Lecomte]

@_lobotomyx @DanielVavra @LarAtLarian Did you try Claude Opus 4.5 ?