Gary Marcus

Rank 33 of 47
|
Score -1
6/20/2024, 4:55:46 PM
X
In reply to:
XINYI LI
@TinaXinyi
·
151d
if most score on those benchmark are around 90% or more, you cant tell if it is convergence or just benchmark design problem
Gary Marcus
@GaryMarcus
·
151d
Claude 3.5 Sonnet hot take: Yet another model, roughly in same ballpark as many others.

The field spent over $50B last year trying to decisively beat GPT-4, but so far what see evidence for is convergence, rather than continued exponential growth.

The statement by @TinaXinyi and the reply by @TinaXinyi engage in a substantive discussion about the performance benchmarks of AI models, specifically Claude 3.5 Sonnet, in comparison to other models. This discussion contributes to the broader public discourse on the progress and evaluation of AI technologies.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement and reply do not cause harm and are focused on a technical discussion about AI benchmarks. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The discussion promotes understanding by addressing the nuances of AI model performance and benchmark design. [+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 reply engages constructively with the original statement, providing a counterpoint without resorting to personal attacks. [+1]