Gary Marcus

Rank 14 of 47
|
Score 89
3/16/2024, 3:37:02 AM
X
In reply to:
Reid Southen
@Rahll
·
570d
@EMostaque Hey Emad, you literally had me blocked until you decided to respond to this. I also never said it was 100,000 to 1 lossless, YOU said that. If you want to talk good faith, let's start with you.

And in terms of compression, how do you respond to this?

https://spectrum.ieee.org/midjou…
Emad acc/acc
@EMostaque
·
570d
The thing that gets me about many of the folk anti-gen AI is the dishonesty of some of their statements.

Like this guy knows this isn't compression as he is smart, but tries to make out like we can losslessly compress 100,000 to 1.

Make good faith points as this is important.
Reid Southen
@Rahll
·
570d
AI proponents keep making the argument that "AI models aren't compression", even though the CEO of Stability AI just flat out says it is 🤷

https://perilli.com/ai/emad-most…

The statement is part of a technical discussion regarding the nature of AI-generated content and whether it constitutes a form of compression. The tone is analytical and aims to clarify the nature of the information retained by AI models. The statement does not appear to be harmful, disrespectful, or unconstructive. It contributes to a dialogue about the technical aspects of AI and its implications.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm as it is a technical observation. [+1]
  2. 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 promotes understanding by contributing to a technical discussion. [+1]
  3. Principle 5:
    I will acknowledge and correct my mistakes.
    The statement is part of a correction or clarification in an ongoing conversation. [+1]