Ever since ChatGPT became available, we’ve seen all kinds of reactions, especially in the mainstream media.
Artificial intelligence explained to my father
So when Michel Sardou came to BFM TV, we showed him how the tool could write a song. The singer’s reaction left no doubt about his feelings. But, among the viewers of the show, there was my father.
He was surprised by the mediocrity of the text generated, which allowed me to explain to him how this type of tool worked.
In the world of Black SEO, we know AI tools very well: we use them every day. We have dozens of content writing services, which boast of being able to generate relevant, well-written and well-calibrated texts for SEO.
Having tested many of them, I might as well tell you that the reality is very different from the marketing discourse. Most are pretty mediocre. ChatGPT is more qualitative, even if the writing quality is not optimal. Moreover, as with all content that you have not written yourself, you must always check that the information contained is up to date. Rest assured: it’s not tomorrow that real journalists will be replaced by artificial intelligence.
The Brotherhood of Whiners
Not a day goes by without a professional body being moved by its possible disappearance because of the existence of this tool. This is making a lot of case for what is only a tool. Activities that have a high advice component will not disappear. On the other hand, creating a tool that will help businesses find answers more easily can only be beneficial. After all, if people have chosen to have a profession with an important part of human contact, it is not to spend more time seeking information than delivering it. When I go to see a lawyer, it’s not for him to tell me all the doctrine — which I can find on my own. When I talk to my accountant, it’s not so that he can get me the official bulletin of public finances, but so that he can tell me “no, you don’t do that” or “yes, you can do that , but pay attention to such points of your situation”.
What about learning? Teachers are worried that the tool could be used by students, who would be tempted to cheat on homework. If the fear is legitimate, it may also be a sign that the system needs to be rethought. When I think back to my law classes, I remember the compulsory cramming. I stupidly learned what I was asked to learn, especially for my course in international relations. I was lucky enough to type very quickly, so I had written down my teacher’s lecture, word for word. Orally, I spit out the course word for word and as I like to clown, I presented my oral on the job of ambassador, in the form of a sketch. But, I don’t remember the course. In the same way, it was only when I took my first finance bill—well after my studies—that I understood the public finance course.
If, from my first year of law, my professors had told me “to understand constitutional law and public finance, follow the finance bill in Parliament”, not only would I have done it, but in addition, I would have succeeded in validating the two subjects without difficulty. Instead, I had classes that were too theoretical, too abstract. It was not justified, because fundamentally, constitutional law and public finance are matters of law that move very little. Of course, we have case law, we have changes to the rules of the National Assembly and the Senate and even constitutional reforms. But, in relation to social law (labour law, social security law) or civil law (family law, contract law), we do not have a revolution every morning. ChatGPT wouldn’t have changed anything in my time, but if I had been given the right tools, I would have understood faster.
Besides, not all subjects lend themselves to cheating with ChatGPT. I played with it for Parliament. I don’t know if it was intentional to want to upset the French Senate, but making it a purely advisory body was not very nice. From this point of view, we have the impression of attending the same debates as at the time of Wikipedia, when the teachers said that the students would use it to cheat. We don’t want to offend anyone, but even in the days of encyclopedias, we used them to fulfill our duties. Trying to cut short homework is a sport collectively practiced by children around the world.
Tech needs to get smarter
There may be a profession that will undergo a revolution: technical support. Last week, I opened a ticket to a company’s IT support. I explained my approach, what I wanted to do and in essence, I asked if it was possible. After five days, the support deigned to answer me saying that it required technical knowledge and that I had to ask someone who had the skills. ChatGPT had given me an almost identical answer. Except that it took him five seconds and before calling me an idiot, he explained to me how to achieve the desired result.
I’ve always been struck by the lack of empathy and humility in the French tech world. We find ourselves confronted with a problem, we ask the question and we get a “go read the documentation”, documentation that is not suitable for someone who is just starting out. ChatGPT is a better pedagogue and here is a concrete example.
In the SANS newsletter, the institute reported on a clever guy who had used the tool to code malicious software. I have a deadlock with Python, whose logic I’m having trouble understanding. I saw hundreds of YouTube tutorials, several people tried to explain to me, I read books, nothing to do. Why not try with ChatGPT, taking a concrete case? So I asked him to do something, in Python. He gave me the code, explaining it to me almost line by line, commenting on it. He gave me the walkthrough, start to finish. An hour later, I understood, I had a working script and I wanted to do other things with Python. Each time I encountered an error, he explained the probable causes to me, to the point that now I have acquired new reflexes.
It’s a simple tool. It’s just code. Who succeeded where all the others, who wanted to be pedagogues, failed. It made me want to spend time in front of my Linux terminal again. Maybe this tool will give a slap in the face to those who thought they were kings of the world for years and they will finally come back to Earth. The debate on ChatGPT is all the more interesting as it will put the human at the center. We will probably integrate that simply telling people “you install such and such a tool and it works” — implied, if it doesn’t work, you’re stupid — that’s not enough. It was never enough and we all have examples of crap user interfaces, lunar documentation or tutorials where the videographer shows his head more often than his screen.
If only because it will calm each other’s egos, we want to say thank you. As for me, I go back to doing fun things with Python.