The place of artificial intelligence today in our world. Should I brake or open the valves? This is the question of the day with the sociologist Jean Viard. Is ethical technology possible?
This week, the CNIL, the National Commission for Computing and Liberties, the policeman of personal data in a way, announced that it had received the first two complaints in France against the much talked about chatbot, the robot American chatGPT (Generative Pre-trained Transformer). We are not anti-tech, but we want ethical technology, explained the lawyer at the origin of one of these complaints. The decryption of sociologist Jean Viard.
franceinfo: An ethical technology Jean Viard, how to answer all these questions and complaints that arise?
John Viard: I do not believe that innovations can really be slowed down. We can supervise them, we can debate how we monitor them. But when a tool has such power and brings so many things, when we say: it will eliminate 300,000 jobs or huge numbers, yes, that means that people will be freed from repetitive tasks, especially in law, for example . Because this tool is going to tell you all the right about a lawsuit, or go two pages to tell you all the lawsuits that there have been on the same subject, so it’s a very effective tool, but it’s a tool who does not tell the truth. It is not interested in the question of truth, and it is a tool that does not invent. Only man can invent, that’s the role of artists, intellectuals, etc., so you mustn’t play the wrong game. But it’s true that he goes so fast that he’s so strong he can be scary.
So we are in this situation. How are we going to build accountability? Who is responsible for people using GPT? They will believe it to be true. I read on the internet and found this to be fair enough: “It’s going to be April 1 every day”. That is to say that on April 1, when we look at our newspapers, what we are looking for is false information. Because, basically, you’re going to have to learn to do that every day. I would say that was already a bit like that with the conspiratorial world, the digital world, but it is a question that is absolutely essential. It is that indeed it will be necessary to reassure the truths, to check the sources.
So there is a huge job in knowing who is going to be responsible for the information that is given. Because if you are given a text with a false source, who is responsible? When I publish a book, if my source is false, I agree to be told. This is the democratic debate.
Jean Viard, you are an author, you are also an editor at Aube editions, you receive a lot of texts from authors. You took an initiative precisely, to check that there were not forgers, in a way, who submit to you texts written by ChatGPT?
Yes, we have already discussed it, and we are looking for a start-up that allows us to pass the texts to him to see if it is actually from ChatGPT, all or part, not necessarily all. But I think it’s the same question that will arise in the universities, that is to say that there will be detectors. Somewhere, there is a bit of a “cat candle” side in fact, that is to say, both it’s exciting, and at the same time, the connections, all that is a bit always the same . I would say: there is a ChatGPT language. And indeed, we decided to find out how to buy software, if it is accessible, to pass the manuscripts through it, to avoid being drawn into manuscripts which would be, in part, fake manuscripts.
So, in the face of what some consider to be excesses, you spoke about the substance. There is also the use of personal data. Italy has decided to block ChatGPT, precisely because the robot would violate the GDPR, the European regulation on this personal data. We will not ban ChatGPT in France. This is what the Minister of Digital, Jean-Noël Barrot, said, who nevertheless shares the observation: “The question posed by Italy is subtle and interesting, even if it gives the wrong answer.”he said…
Europe is ahead of the United States. The United States has no rules in all these areas, yet it is still the United States that carries this, including in ChatGPT, most of the documents that feed the answers to our questions come from the Anglo-Saxon world. And indeed, we try to fit in, we have to regulate. We are in a new industrial revolution. We are in 1848, at the time of the transformation of England into an industrial country, it is the same style of rupture of societies. And we entered the green industry. We have entered a society that will produce a low-carbon world and which is an exciting and extremely dangerous new adventure.
So you have to say that it is also an extraordinarily powerful tool for this fight. But it’s like the printing press, when the printing press became democratized, and the Bible actually became accessible to everyone, and that led to the religious ruptures that we know, there are always people who say no, the printing press must be stopped! No, the printing press must not be stopped, the populations must be educated. Let’s learn to play with it, learn to use it and above all, don’t run away, you know you never win a fight when you run away…