I've made no particular secret of the fact that I am not a fan of large language models: they're overhyped, under deliver and were built off the back of a ripping off of others work at a scale that (for anyone else) would have resulted in jail and/or bankruptcy (even before the more incontrovertible acts of copyright infringement are considered). But, I also work in a industry which is going through a phase of significant AI use (even if a lot of companies do come to regret being over-eager)....