Gary Marcus

Rank 25 of 47
|
Score 40
Ethan Mollick
@emollick
·
434d
Here is why I am so doubtful about the "talk-to-your-data" use of AI.

This is Google NotebookLM, a cool tool that lets you use AI on data sources. Even though the document search retrieves the right information (it is Google, after all), the LLM answer has subtle hallucinations.
1/15/2025, 3:30:17 PM
X
In reply to:
Ned Nguyen
@NedNguyen
·
26d
@GaryMarcus Hallucinations issues have been mitigated quite a lot since mid 2024. I used to be very skeptical of LLM IR but in the past quarter, I can trust notebooklm to give me very reliable information about large documentation corpus.
Gary Marcus
@GaryMarcus
·
26d
@NedNguyen I prefer that my informational retrieval mechanisms not hallucinate.
Ned Nguyen
@NedNguyen
·
26d
@GaryMarcus LLM alone won't give us true human-like intelligence, but it's a very important milestone in AI research, don't you think?

At minimal, it enables way better information retrieval capabilities than TF-IDF
Gary Marcus
@GaryMarcus
·
27d
In a galaxy far, far away, back in 2001, when people who cared about the nature of intelligence studied things besides LLMs.
Aashish Reddy
@_AashishReddy
·
27d
Went to the end of @GaryMarcus 'The Algebraic Mind' to find a names index, preceded by a foreword, preceded by references, preceded by notes. Was briefly annoyed and perplexed why there would be no actual index, but it was just before! Thought I'd share a subset:

The conversation involves a discussion on the reliability and development of large language models (LLMs) in information retrieval, touching on issues like hallucinations and advancements in AI. The tone is analytical and focuses on the technical aspects of AI development.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The discussion is technical and does not appear to cause harm. It focuses on improvements and challenges in AI, which is constructive. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The conversation promotes understanding of AI advancements and challenges, fostering a deeper comprehension of the technology. [+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 dialogue is respectful and engages in constructive criticism, particularly regarding the reliability of LLMs. [+1]