Chollet Says AI Will Shift From Current Deep Learning Patch
Chollet argues the field applied a temporary deep learning patch now in wide use.
François Chollet stated that to overcome deep learning limits the AI field applied a first patch demonstrated in December 2024 and now ubiquitous. He maintains that a second patch remains inevitable over the longer term. Machine Learning Street Talk posted that Chollet made the same prediction in early 2024 and that recent developments align with it. The posts present these statements as Chollet's own assessment rather than confirmed outcomes.
To address the limits of deep learning and avoid stalling, the field of AI started by applying patch (1), which started being demoed 9 months later in December 2024 and has now become completely ubiquitous. However, long term, it is simply inevitable that AI will move to patch…
There are essentially two main options to remedy this: 1. Find ways to perform active inference, so that the model adapts its learned program in contact with a new data distribution at test time. Would likely lead to some meaningful progress, but it isn't the ultimate…
Reactions
1 postML 
Machine Learning Street Talk@MLStreetTalkChollet called it in early 2024, he's still right 😀
- likes: 49
- replies: 5
- bookmarks: 31
- reposts: 2