Research
For a very long time most high-performing AI models were end-to-end neural models; vector input -> vector output, with only ultra-thin symbo…
For a very long time most high-performing AI models were end-to-end neural models; vector input -> vector output, with only ultra-thin symbolic preprocessing and postprocessing layers (e.g. label deco
For a very long time most high-performing AI models were end-to-end neural models; vector input -> vector output, with only ultra-thin symbolic preprocessing and postprocessing layers (e.g. label decoding). For many, it seemed that moving more and more logic to the end-to-end neural model was the way of the future. "Differentiable programming". But what we have now is heavy neurosymbolic systems where the model itself is symbolic.
Related
- Symbolic learning is not a replacement for coding agents, it's a replacement for gradient descent & NNs: a low-level, completely general, ex…
- Even right now, many workflows are morphing into LRM-guided harnessess that manipulate symbolic programs. Which is a crude, but currently-ac…
- The most complex phenomena arise from scalable recombination of very simple rules. Whether it's galaxies, chips, or neural networks, if you …
Source: Francois Chollet (X) | 2026-08-06