Yann LeCun, an A.I. Pioneer, Warns the Tech 'Herd' Could Hit a Dead End
nytimes.com>He argues that the technology industry will eventually hit a dead end in its A.I. development
I'm not sure it'll be a dead end. Probably not straight ahead but they may branch off into variations.
Already large language models have modified into multimodal models that can work with images and video.
Something is probably needed so model weights can be updated from experience rather than just pretraining but that doesn't necessarily mean scrapping it all and staring over.
Yann Lecun is a legend, no doubt, but, his critique is starting to become outdated. He wants the whole world to slow down and wait for his preferred paradigm, but everyone is out shipping instead. We have LLMs that pass BAR exams now and do multi-step "reasoning".
I am an LLM reasoning skeptic too, but the problem is, he's dismissing real measurable progress while not proving his own alternative approach.
He's starting to sound the like stochastic parrot crowd, shaking fists from the side line while everyone and their dog throws these systems into a pot and sees what they can cook up.
Why does he spend so much time in the media to decry the current AI center of gravity?
Instead, he should spend long days working to prove his contrarian hypothesis.
Maybe people come and ask him all sorts of questions, and he simply wants to be nice.