Gary Marcus

Rank 10 of 47
|
Score 110
10/3/2024, 9:03:21 PM
X
In reply to:
Carl Pham
@CarlPham9000
·
296d
@GaryMarcus It's seems unlikely the networks themselves are too impoverished -- aren't they Turing complete? What is much more believable is that the *training regime* is too blunt, that to actually train it effectively would take all the data (and time) since the Big Bang.
Gary Marcus
@GaryMarcus
·
296d
I think this is totally wrong, and that (a) we will never understand the brain without having a better handle on why brains have so many cell types and (b) that neural networks as we currently build them are too impoverished to get to general intelligence

But I would also say…
AGIHound
@TrueAIHound
·
296d
The large number of cell types is not something we need to worry about too much imo. Biological brains develop from tiny embryos. The various cell types are needed to achieve the correct innate synaptic connections during the growing phase. All neurons process events (spikes).

The statement engages in a technical discussion about the limitations of neural networks and their training regimes in achieving general intelligence. It critiques the capacity of neural networks to extrapolate from data, suggesting they are 'impoverished' in this regard. The tone is analytical and speculative, contributing to an ongoing debate in AI and neuroscience.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm; it is a technical critique.
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by engaging in a discussion about AI limitations. [+1]
  3. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The statement engages in constructive dialogue, responding to a previous point with a counterargument. [+1]