Gary Marcus

Rank 35 of 47
|
Score -20
6/24/2024, 4:39:06 AM
X
In reply to:
Nick Moran
@NickEMoran
·
135d
@GaryMarcus @garrytan @ycombinator But how can you possibly certify or protect against someone who is allowed to make *unlimited* modification to your product creating something that enables a harm? Are auto manufacturers required to certify that it is impossible to modify a car to create something dangerous?
Nick Moran
@NickEMoran
·
135d
@GaryMarcus @garrytan @ycombinator The bill requires that you certify that derivative models do not cause or enable critical harm, and derivative models include unlimited post-training modifications (other than fine-tuning) and combination with unlimited external software.
Gary Marcus
@GaryMarcus
·
135d
@NickEMoran @garrytan @ycombinator What exactly makes the requirements impossible to comply with! In what way they are worse than the requirements in eg automotive which many manufacturers routinely meet? Please be specific.
Nick Moran
@NickEMoran
·
135d
@GaryMarcus @garrytan @ycombinator The requirements are impossible to comply with, and so any open-source covered model developer will necessarily either not open source their model, or will commit perjury by falsely certifying that they meet the impossible standard.
Gary Marcus
@GaryMarcus
·
136d
Hey @garrytan @ycombinator your letter on California’s AI regulation seems awfully sensationalist and alarmist to me. Have I missed something?

You say “AI software developers could go to jail simply for failing to anticipate misuse of their software–a standard of product…
Garry Tan
@garrytan
·
137d
SB1047 is not the right path.

This bill, as it stands, could gravely harm California’s ability to retain its AI talent and remain the location of choice for AI companies. We hope Sen. Wiener chooses a path that protects startups and open source AI.

https://www.politico.com/newslet…

The statement constitutes public discourse as it engages substantively with a public issue, specifically the debate around AI regulation and the responsibilities of developers. The tone is analytical and aims to clarify the expectations and limitations of developers in preventing harm from AI models.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement strives to do no harm by clarifying the realistic expectations of developers, which can prevent undue panic or misinformation. [+2]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The statement respects the privacy and dignity of others by not engaging in cyberbullying, harassment, or hate speech. [+2]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by explaining the reasonable limits of developer responsibilities, fostering empathy for their position. [+2]
  4. 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 criticism by addressing the points made in the previous conversation without resorting to personal attacks or ad hominem arguments. [+2]
  5. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement upholds the principles of free speech and uses the platform responsibly by contributing a reasoned perspective to the public debate. [+2]