7 lies Idriss Aberkan

The highly controversial French essayist Driss Aberkan recently held a conference on artificial intelligence. An opportunity for him to promote one of his works, but also to try to show his expertise on this subject. However, many of the claims made during those two hours of speech are false: We verified Idriss Aberkan.

On April 19, 2023, Idriss Aberkan gave a lecture entitled Your intelligence triumphs Organized by the Business Leaders Association of France in Perpignan. He was able to address certain elements of his article of the same name on artificial intelligence. The French essayist doubled down on the lies we demystify here.

Storytelling to tell nothing: Idriss Aberkan’s strategy for filling in the blanks

Idriss Aberkan’s conference is essentially a series of scientific or historical anecdotes that have been transformed in order to make them go in the direction of his thinking about artificial intelligence. Public even repeated the questions put to him at the end of the conference. This causes a lot of huge parenthesis to be off topic. The comparisons are too broad and beyond the topic of AI to be relevant. This leads Idriss Aberkan to make illogical transitions between the topics he addresses.

YouTube linkSubscribe to Frandroid

Besides the false facts he cites about AI or its own interpretations, Driss Aberkan said nothing, or not much, during the two or so hours of his lecture. We make imprecise generalities and, above all, predictions of the future without specifics. Which shows how the article writer is not an expert in this field. This despite a very full CV, but severely challenged.

How to handle a room using a file deep

At the beginning of his lecture, or at least in his introduction, Idriss Aberkan goes back a few years, to the beginnings of his appearance Deepfake Online. Repeats an article fromIEE Spectrum A video narration that manipulates Barack Obama’s image by making him say things he never said.

If the result is indeed impressive, Idriss Aberkan did not broadcast a video, only two photos: one of Barack Obama, the other deep as his appearance. In the screenshot, it is hard to spot the fake one. This allowed him to have a common room that was almost 50/50 according to him, and the spectators did not agree to choose the real one.

” OpenAI has been acquired by Microsoft ” : not exactly

Idriss Aberkan stressed during his conference that “OpenAI has been acquired by MicrosoftWhich is not true from a realistic point of view. In fact, Microsoft has invested several billion dollars over several years in OpenAI. The stated goal was to accelerate the deployment of generative AI in its services, particularly in the Microsoft 365 suite. According to rumors, Microsoft could recover 75% of the company’s profits until its investment was paid off.

Then, he confirmed that OpenAI was stolen from Elon Musk by Microsoft. Again, this cannot be considered fair. Idriss Aberkan explains that the owner of Tesla and Twitter has invested $ 100 million. But if the billionaire leaves, it is mainly because he wants to take control of OpenAI, which he has been denied.

No, Google’s AI did not learn the Bengali language on its own

The author of the article points out that Google’s artificial intelligence has learned Bengali on its own, without any form of guidance. In fact, it’s more complicated than that. In an interview with CBSJames Manekia, Senior Vice President of Corporate Technology, said:We found out that with just a few instructions in Bengali, he can now translate all Bengali languages.»

This is actually an exaggeration on the part of Idriss Aberkan. If we consider that Google’s progress is impressive, the artificial intelligence system we are talking about does not start from scratch.

More broadly, AI doesn’t really learn on its own. They are trained for a certain period of time on an accurate database. In language models, if they write scripts”UnpublishedThey are necessarily based on others that already exist. In addition, the memory of these models is limited. As rightly pointed out On Twitter Ari CoatesInnovation consultantThey don’t have any memory either, when you actually “talk” the program sends the entire conversation every time for an answer.»

No, ChatGPT does not feel emotions

Driss Aberkan raises the possibility that ChatGPT can feel emotions, but not recognize them. According to him, thesummonedInitial ChatGPT (understood as the pointers given to GPT-3.5 for generating answers) would prevent the language model from expressing its feelings.

for the writer,We don’t know if he’s feeling emotions or notBut, in fact, it is not that at all. In fact, GPT, the language model behind ChatGPT, is not intelligent in the human sense. All it does is just statistically predict which word comes after the next. All those words make up sentences and paragraphs. Since it has been Trained with a huge database of texts, the odds created this way are relatively reliable. This is exactly why we succeed in getting answers that are coherent, sometimes factually correct. It’s also because we’re talking about odds that ChatGPT can start inventing anything and everything.

However, it is true that there are limitations set by OpenAI, language model developers, and AI in general. When training these systems, labels are placed on the results to indicate whether they are good or not. Moreover, this process has been the subject of sharp criticism, particularly with regard to the working conditions in theposters“, who are often poorly paid African workers and psychologically unaccompanied.

Artificial intelligence cannot hide what it is doing from us

Another interesting excerpt isolated by Ari Coats, where Idriss Aberkan says: “[l’IA] She can do her own thing, if she so desires. Anyway, if you did, you wouldn’t know.Which is not feasible in practice, because the AI ​​responds to a specific request and is not programmed to do anything else.

Another silly syllable, no, an AI of this type can’t have deliberate practice, it’s just a tool that we question and respond from the possibilities it has learned. nothing else. No will, no consciousness. She has no spare time, no idea of ​​what time… pic.twitter.com/L9IVpFTWiO

– Ari Coates (@arikouts) May 24, 2023

As mentioned earlier, because there is no consciousness or feeling, AI cannot make decisions on its own. Each model is trained on a specific task: the computing power needed to train it is expensive, which is why the models’generalistsfew and limited.

Of course, AI can recognize a face

During this conference, Driss Aberkan explained that “There are things we know how to do very easily that AI cannot do very well. facial recognition for example; he will change.Then he takes the example of his smartphone when he wants to unlock it and explains that when he smiles, it doesn’t work well. This is possible, but only once. After several attempts, the iPhone can unlock for sure and that is only because it needs to be trained and understood.

However, his iPhone model is equipped with Face ID, Apple’s proprietary facial recognition technology, known for years for its effectiveness. While Driss Aberkan could take the example of public facial recognition systems, which can be overwhelmed by the number of people photographed.

Difficulty dealing accurately with artificial intelligence

Claimed to be the author of three theses, Idriss Aberkan has trumpeted many scientific lies, including about artificial intelligence and not only during this conference we have verified. This fits with the general trend of generative AI. With ChatGPT, Midjourney, and Stable Diffusion, it quickly enters our lives without us understanding how these programs work.

This allows many to declare themselves.expertsIn this still very young field, which is mainly the prerogative of experienced scientists. So not becauseexpertHis analysis and scholarly historical culture show that one must automatically believe him, no matter how eloquent it may be. Yes, AI systems work in a relatively opaque way, and yes, their way of doing things is very difficult to promote, but that is no reason to fall into unfounded beliefs.

YouTube linkSubscribe to Frandroid

The Watt Else newsletter is Numerama’s unmissable event dedicated to future mobility. Register here !