Gary Marcus

Rank 16 of 47
|
Score 76
4/21/2024, 4:22:07 PM
X
In reply to:
Ahmed Mekallach
@Ahmekal
·
424d
@GaryMarcus @OpenAI I build reasoning systems, so I'm living the results.

I can confirm GPT4 transformers in a network produce intended output with simple and thought-out prompt engineering.
Gary Marcus
@GaryMarcus
·
424d
@Ahmekal @OpenAI oh come on. are you saying that GPT-4 never makes errors like these? or just that on one particular trial it happened not to make this specific error?
Ahmed Mekallach
@Ahmekal
·
424d
@GaryMarcus Just use a real transformer like @OpenAI GPT4 for reasoning tasks.

Stop wasting time on anything else. Spend the money

When open models get to GPT4 level, and they will, then make the switch to them.
Gary Marcus
@GaryMarcus
·
424d
Unless someone else comes up with a better term for the trifecta below, I am going to call it “automated, clueless ass-kissing” and welcome a GIF or meme to illustrate it.
Gary Marcus
@GaryMarcus
·
424d
Help! Is there a term for this particular LLM trifecta of discomprehension, overconfidence and obsequiousness? 🙏

The statement '@Ahmekal @OpenAI if you think that is invariably true, the laugh is on you.' is a response in a technical discussion about the reliability and performance of AI systems, specifically GPT-4. The tone is somewhat dismissive and sarcastic, aiming to challenge the previous assertion about the infallibility of AI systems in producing intended outputs.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm but uses a dismissive tone which could be seen as slightly harmful in a professional or technical discussion. [-1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement respects the dignity of others to a degree but could be more constructive in fostering a respectful dialogue. [-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 lacks a compassionate or empathetic approach, which could have facilitated a more productive discussion. [-1]
  4. Principle 5:
    I will acknowledge and correct my mistakes.
    The statement does not acknowledge any mistakes but challenges the assertions made by others, which is a form of engagement in dialogue, though not necessarily constructive.