The 🗝️ fact: You can’t understand contemporary AI unless you can appreciate the difference between a model (that is part of a system) and a system as a whole.
If a system as a whole includes both neural networks and symbols, it is by definition neurosymbolic.
To think otherwise is like thinking that an engine IS a car, rather than a component of a car. You can’t make a car out of engines alone. and you can’t (in most cases) make robust, reliable AI systems out of neural networks alone.
Instead, in many domains you get better results by *integrating* both neural networks and symbols. (Math, where e.g., Lean checks and sometimes rejects the work of LLMs is a good example of this; so is Claude Code).
That was literally the point of my 2001 book The Algebraic Mind and has been a central point throughout my career.
For a long time, people like @GeoffreyHinton argued otherwise, suggesting that all we needed were neural networks by themselves, and that symbols were an archaic waste of time. Hinton likened symbols to “aether”, as if they were complete scientific mistake.
He was wrong. Systems like Claude Code depend heavily on combining neural networks with things like rules, code interpreters, functions, conditionals and loops - all symbolic constructs drawn from classical AI. Math systems use automated theorem provers, which flow directly from Newell and Simon’s symbolic work in the 1950s.
Symbols weren’t a mistake; they were vital, just as neural networks are. Scaling helped, but by itself it wasn’t enough; symbolic tools and code interpreters and harnesses are now a central part of most working systems.
Neurosymbolic AI has won, overwhelmingly; the pure neural networks that the dogmatists like Hinton used to argue for have been almost entirely displaced by neurosymbolic hybrids.
As @fchollet has also noted, to pretend otherwise is just silly.
Today, the promises of transhumanism and posthumanism are multiplying. These are ideological currents according to which, sooner or later, technology will eliminate all disease and pain, as well as aging and even death itself. Precisely because scientific progress is a supreme achievement of human intelligence, it must not be confused with a privilege reserved for millionaires seeking immortality, while entire peoples suffer from poverty and oppression. In the face of this mirage, which promotes the illusion of human beings perfected by cybernetics and biotechnology, Christian anthropology reminds us that a limitation is not a manufacturing defect, but rather the very condition of human grandeur. #ApostolicJourney #Peru #MagnificaHumanitas https://t.co/wOXDtgnSi5
hot take: "Pathfinding" is a terrible metaphor to understand how reasoning models work. A more useful metaphor, IMHO, is "learning prompt augmentations." There is no "token space search"--as in concatenating tokens up to construct paths towards solutions. The only search is a "Generate-test" variety where full trajectories are generated and the ones ending up in non-solutions are rejected. The GRPO part implicitly boosts the trajectories ending in correct solutions, thus helping the system learn the right prompt augmentations to mutter to itself.
My impression, based on using LLMs, is that their effectiveness in solving mathematical problems depends heavily on the quality and untapped potential of the existing literature underlying those problems.
If that is correct, I wonder what will happen once that potential has been exhausted, especially if mathematicians become overly reliant on LLM assistants and the literature is not replenished by genuinely new, sufficiently deep insights.
In other words, I’m not entirely sure we aren’t entering a creative rabbit hole.
@_akpiper I have a somewhat different view of these matters. I think AI will allow us to ask new questions and do new kinds of analysis, especially of form. See, e.g. From LLM mechanisms to ring-composition: A conversation with Claude 3.5, https://t.co/tG2Lk6vJb0