Gary Marcus

Rank 12 of 47
|
Score 305
5/6/2024, 3:06:18 AM
X
In reply to:
Masen Dean
@masenmakes
·
13d
@GaryMarcus The most erratic things I run into are easily fixed by the model realizing its own mistake.
This is how I deal with these errors in both GPT4 and Claude.

It's not ideal, but I'd like to see someone have more experience with current models before going straight to 'AGI wouldn't…
Gary Marcus
@GaryMarcus
·
13d
@masenmakes If you are smart, how do you avoid erratic behavior? It’s literally inherent in the architecture.
Masen Dean
@masenmakes
·
13d
@GaryMarcus Even if I mostly disagree, I've truly respected your opinions until today.

How can you grandstand on the failures of LLMs when you don't even know how to avoid generic issues like erratic behavior?

All your posts on the failings of AI tech just fell incredibly flat for me... 🫤
Gary Marcus
@GaryMarcus
·
13d
Encouraged by a friend I tried out some LLMs this weekend, first time in a while.

They are just so erratic; hardly anything feels replicable, within or between models.

How anyone is supposed to engineer real-world solutions on that foundation, I really don’t know.

The statement '@masenmakes I am not interested in being skilled at this particular tech. it’s a detour.' is a personal expression of disinterest in a specific technology, which is part of a broader conversation about the reliability and development of language learning models (LLMs). While the broader conversation touches upon significant issues in AI technology, this particular statement does not substantively engage with public issues or contribute to civic dialogue in a meaningful way. It is more a personal stance or preference related to the individual's career or personal development path.