Gary Marcus

Rank 15 of 47
|
Score 76
6/16/2024, 4:40:47 PM
X
In reply to:
Anthony Rinaldi
@anthonyprinaldi
·
511d
@GaryMarcus Which systems can we rely on for solving these abstract (and hierarchical) problems? Is semi-symbolic the way?
Gary Marcus
@GaryMarcus
·
511d
What I think many people don’t get is that the problems generative AI that we are seeing now aren’t new; they aren’t random. They are *persistent*, e.g., neural networks have NEVER been reliable at negation, NEVER been reliable at compositional structure, and NEVER been good at…
T J
@TJ09299872
·
511d
@iancc_wip @GaryMarcus GenAI is pretty good at synthesizing concepts with a vast amount of training and therefore latent representation, but it's also trivial to come up with examples they cannot produce

The statement 'Neurosymbolic is a necessary step' is part of a broader discussion on the limitations and potential solutions for generative AI systems, particularly in handling abstract and hierarchical problems. The conversation involves multiple participants discussing the reliability and common sense understanding of AI models, with references to specific examples and benchmarks.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm and is a constructive contribution to the discussion on improving AI systems. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by suggesting a specific approach (neurosymbolic) as a necessary step, which can foster further discussion and exploration of this idea. [+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 by responding directly to a question about potential solutions for AI limitations, without resorting to personal attacks or ad hominem arguments. [+1]