Category: AI, Computation, and Technology

Google is your friend. If you have trouble understanding then ask OpenAI or Grok for an explanation.

That’s an oversimplification of this early stage, better called synthesis. Once permutive, recursive, and adversarial, it’s truly thinking.

Q: CURT: Liquid Neuron? (Vienna University > MIT > Startup)
Yes. I am aware of the general strategy of using differential equations for each node and the benefits that would accrue to it. I’m still in the camp that these approaches, while more efficient (much) are too centralized to reduce the cost of adaptation – and that only neuromorphic computing can achieve that reduction of costs and time. That said, it’s possible that there is some solution that is so efficient that existing tech can do the job just as well. But every edge tech I’ve seen from chip makers tells me we will probably reach convergence with neuromorphic computing if we don’t reach replacement of existing tech with it.
The primary benefit of the Liquid Neuron approach is the results are open to introspection, where existing networks are not. IMO while I would prefer neuromorphic hardware, it may be true that local calculation of these differential equations is a sufficient replacement for the density and complexity of dendritic precision.

Links:
1) https://t.co/G1JCfZmPze
2) https://t.co/oAX7JYQUxY
3) https://t.co/ypH8YXipFz

Reply addressees: @SurragoMichael @grok