Gary Marcus

Rank 37 of 47
|
Score -82
8/2/2024, 10:58:50 PM
X
In reply to:
Phil Libin
@plibin
·
91d
@VrDevBrad @GaryMarcus Yes, it’s obviously low quality bullshit. I think this was @GaryMarcus’s point. Funny bullshit.
Bradley Austin Davis
@VrDevBrad
·
91d
@plibin @GaryMarcus Spending less than 5 minutes trying to search for its provenance reveals it's a poorly photoshopped version of the image used here: https://www.linkedin.com/pulse/g…

The bullshit here is clearly coming from Planet VPN, not "AI".
Phil Libin
@plibin
·
91d
@VrDevBrad @GaryMarcus True. Looks like you’ve found an inconsistency in something related to AI.
Bradley Austin Davis
@VrDevBrad
·
91d
@plibin @GaryMarcus Settle down. By your reasoning, the 75 billion listed for GPT-4 would be over 7 millennia of GPU production, so it shouldn't exist.

The labels are probably referring to GPU-hours, or GPU-cores or GPU-hour-cores.

A single modern GPU can provide ~50m GPU-core-hours over a year.
Phil Libin
@plibin
·
92d
@GaryMarcus Ten trillion GPUs you say? That’s about a million years worth of Nvidia’s annual production of data center GPUs. Sounds like GPT-5 might not ship this fall. https://twitter.com/plibin/statu…
Gary Marcus
@GaryMarcus
·
92d
There is a critical error of logic here.

GPT-5 surely will use far more computational resources than GPT-4. 𝘽𝙪𝙩 𝙞𝙩 𝙞𝙨 𝙨𝙩𝙞𝙡𝙡 𝙖𝙣 𝙤𝙥𝙚𝙣 𝙚𝙢𝙥𝙞𝙧𝙞𝙘𝙖𝙡 𝙦𝙪𝙚𝙨𝙩𝙞𝙤𝙣 𝙝𝙤𝙬 𝙚𝙣𝙝𝙖𝙣𝙘𝙚𝙙 𝙞𝙩𝙨 𝙘𝙖𝙥𝙖𝙗𝙞𝙡𝙞𝙩𝙞𝙚𝙨 𝙖𝙣𝙙 𝙥𝙚𝙧𝙛𝙤𝙧𝙢𝙖𝙣𝙘𝙚…

The statement 'Correctamundo. (Funny but also representative a bunch of similar posts.)' is a reply in a conversation about the computational resources required for AI models like GPT-4 and GPT-5. The conversation involves multiple participants discussing the validity and implications of claims about AI capabilities. This qualifies as public discourse as it engages with a public issue related to AI technology and its development.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm with words and actions. It is a lighthearted agreement with a previous comment. [+1]
  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 and does not engage in cyberbullying, harassment, or hate speech. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement does not actively promote understanding, empathy, or compassion, but it also does not detract from these values.
  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 does not engage in constructive criticism or dialogue but rather agrees with a previous comment in a humorous manner. It does not involve personal attacks or ad hominem arguments.
  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 and with integrity, as it contributes to the ongoing discussion without causing harm or spreading misinformation. [+1]