I'm wondering if the CEO just understands their clients will buy more products if they have "AI" in the title, even if it's pointless.

Your job isn't to make good software, it's to make profitable software.

This was an internal tool, so no use in fluffy marketing.

I'm not saying it was a sensible decision, but I think at that level you kinda need to have a few balls in the air. Having a few projects using ML or whatever else is in vogue is defensible, might be a win, might build some expertise, shows you're in touch with the current thing etc.

Yup exactly, sometimes you should try new things. They don't always work out. Although in this case it seemed like an expensive mistake...

I think this was a folly by our director. About a year after I left I learnt that he had "left to pursue other opportunities", which is corporate speak for "he got canned".