Gary Marcus

Rank 19 of 47
|
Score 64
6/23/2025, 3:06:59 AM
X
In reply to:
Lawrence Chan
@justanotherlaw
·
18d
@GaryMarcus But my guess is there is no interpretable, reliable AI paradigm that can outrace LLMs at this point with any reasonable sized research program. If you want to build tractable, transparent alternatives to LLMs, you'll need to work on either getting way more cognitive labor, or
Lawrence Chan
@justanotherlaw
·
18d
@GaryMarcus This step 1) contains your options 2. and 3., while step 3) contains option 4 (and, if you really squint, option 1). My guess is you didn't include anything under step 2) b/c you'd argue there's not much useful cognitive labor to extract anyways.
Lawrence Chan
@justanotherlaw
·
18d
@GaryMarcus Nowadays, I think insofar as there is "a plan" that is workable, it'll looks something like:
1) Buy time for steps 2) and 3), either via coordination or by "winning" the race hard enough.
2) Figure out how to safely get useful cognitive labor with current AIs.
3) Use the
Lawrence Chan
@justanotherlaw
·
18d
@GaryMarcus The problem with "make a new AI paradigm" is not a lack of desire. Many AI academics want this, especially those from an older time when making new insights about intelligence was key to AI progress.

The problem with making a new paradigm is that it turned out to be too hard.
Lawrence Chan
@justanotherlaw
·
18d
@GaryMarcus It's not that AIS people haven't considered working on 4 -- Yoshua Bengio and davidad come to mind here as contemporary efforts, Stuart Russell and Eric Drexler as older examples. And of course a lot of mech interp people in 2022-3. I think MIRI also tried this in 2018-20, but
Lawrence Chan
@justanotherlaw
·
18d
@GaryMarcus Of course, these aren't mutually exclusive or jointly exhaustive -- you might hope for 2 3 4 at once (which is what I'd imagine you'd push for).

The statement is part of a broader discussion on AI safety and the development of new AI paradigms, which constitutes public discourse as it engages with public issues related to technology and safety.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not directly cause harm but suggests a need for more effort in AI safety, aligning with the principle of striving to do no harm. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement does not promote understanding, empathy, or compassion, as it is more of a critique without offering constructive suggestions.
  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 is a critique of current efforts in AI safety but does not engage in personal attacks or ad hominem arguments, maintaining a neutral tone. [+1]