There’s a problem in technology: many people believe everything is on the same curve as transistor density. Most things aren’t. There’s often a sharp trend upward and then a stall. https://mastodon.cloud/@jasongorman/112267865098989336
@carlmalamud Feels like the same thing happened with deep learning. I had this conversation in an interview nearly a decade ago with Yann LeCun for a story on computer vision. Despite his co-creation of and intense work on deep learning, he felt it was already nearing its accuracy limits and worried not enough was being done to find fresh grounds to bridge the last gap. LLMs aren’t really much different in nature.
Add comment