I don't get why there hasn't been an #AIWinter, since the "latest" #MLand #NLProc innovations have consistently failed to deliver on their promise to "revolutionize" AI. I remember hearing that deep learning was supposed to be soooo much better than SVMs, and then transformers and LLMs, now *really*-LLMs. Is it that people are afraid of looking foolish if the latest thing actually lives up to the #AIHype? Is it that the same orgs developing the tech are also the ones in charge of hyping it?