@irenes @unlambda @simonlbn i've mentioned elsewhere in this fastly growing thread: i think that's part of the reason why i joined what has now become a bit of a pile-on. the entire industry has been swept up in this "AI" gold rush and i, just one person, am powerless to stop it from happening and prevent all the damage it's going to cause.

but here comes a change that is both:
1. so poorly planned the methodology used to justify it is laughable,
2. being discussed in a public forum where anyone can join and opposition can be built.

i think for some of us this became a way to exert some amount of pushback against the tide, but it also became the one place where our frustration could be channeled...

@nicuveo

Personally I believe we are rapidly heading toward an AI winter of sorts. So my consolation would be that it should all come crashing down shortly (my personal prediction is a year and a half if not sooner).

My evidence for this is broadly the historical cycle of it (every 20 years), the tech slow down (also a historical sign), and the fact that the last 3 years have seen numerous methods of improving AI reach the apex (e.g. we now train on the whole internet, there isn't a second Internet's worth of data to improve the training set; NVidia's new AI cards are no longer showing workload specific improvements, just general hardware improvements, and so on across dozens of factors)

It may also be wishful thinking on my part.