When I don't get something useful with search terms, I often try to ask AI like ChatGPT:

> One famous case that somewhat aligns with this is the Société Générale trading scandal involving Jérôme Kerviel. Kerviel, a former trader at Société Générale, used his knowledge of the bank's systems to make unauthorized trades worth billions of euros, though it was more complex than just modifying a VB script in Excel.

It's interesting because when I do anything like that I don't look at the language information in the AI response.

Yours is a good example. All I'm looking for is the key names that would enable me to actually get the real information without AI repackaging, which I assume to be flawed.

I agree with you on that. In ChatGPT's answer, for me only names and keywords are relevant, which I can then use to start another search, and I ignore the rest of the “noise”. I was also more trying to show with my post that I occasionally use A.I. when I can only vaguely remember something and thus am not even sure what the right search terms for a classic web search would be. Like humans, ChatGPT is often able to extract the right information from imprecise, anecdotal messages and then provide suitable answers that I can then use for my further search.

He was a middle office worker, in security/admin, where he verified then authorized big trades. He used his excel and VB skills to impress a front office manager and get a place in the front office as a market maker (I think he has been assistant analyst in between the two job, but that isn't relevant).

Société Générale used the same id / email adresses for him after he changed jobs, and since the user management was lacking (and wasn't audited), he was allowed to make huge trades as a MM, and authorize his own trade as an ex-security/admin guy.

I might be wrong about his exact job title in the middle office, but I 100% guaranty you the rest.