Gary Marcus

Rank 14 of 47
|
Score 83
Gary Marcus
@GaryMarcus
·
620d
So many people are confused about the relation between human cognitive errors and LLM hallucinations that I wrote this short explainer:

Humans say things that aren't true for many different reasons
• Sometimes they lie
• Sometimes they misremember things
• Sometimes they fail…
5/13/2024, 9:51:39 PM
X
In reply to:
taoki
@justalexoki
·
598d
@GaryMarcus @CreativeS3lf so do you
Gary Marcus
@GaryMarcus
·
598d
@CreativeS3lf yes but it makes silly errors and hallucinates, same as before…
Abdulmajeed
@CreativeS3lf
·
598d
@GaryMarcus The model process 3 modalities better than all models across various tasks, faster, cheaper, and has high throughput compared to old models. The use cases of this technology are insane. Finally, it's beats all models in the LLM arena. 3 modalities ~> it sees, listens, and gets L!
Gary Marcus
@GaryMarcus
·
598d
GPT-4o hot take:
• The speech synthesis is terrific, reminds me of Google Duplex (which never took off).
but
• If OpenAI had GPT-5, they have would shown it.
• They don’t have GPT-5 after 14 months of trying.
• The most important figure in the blogpost is attached below. And…

The statement '@justalexoki @CreativeS3lf you might learn something if you read this?' is part of a broader conversation about the capabilities and limitations of language models. The tone of the statement can be perceived as slightly condescending, suggesting that the recipient lacks knowledge or understanding that could be gained by reading the provided material. The intent seems to be to encourage reading and possibly to correct or expand the recipient's understanding of the topic.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement could potentially cause harm by making the recipient feel belittled or inadequate, which is not conducive to a respectful discourse. [-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 of the individuals involved and does not disclose any personal information. [+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.
    While the intent might be to promote understanding, the tone used may not effectively encourage constructive dialogue or learning.