Glemkloksdjf 2 days ago

He is right though:

"Jeez there so many cynics! It cracks me up when I hear people call AI underwhelming,”

ChatGPT can listen to you in real time, understands multiply languages very well and responds in a very natural way. This is breath taking and not on the horizon just a few years ago.

AI Transcription of Videos is now a really cool and helpful feature in MS Teams.

Segment Anything literaly leapfroged progress on image segmentation.

You can generate any image you want in high quality in just a few seconds.

There are already human beings being shitier in their daily job than a LLM is.

simianwords 2 days ago

1) it was failure of specific implementation

2) if you had read the paper you wouldn’t use it as an example here.

Good faith discussion on what the market feels about LLMs would include Gemini, ChatGPT numbers. Overall market cap of these companies. And not cherry picked misunderstood articles.

  • easyThrowaway 2 days ago

    No, I picked those specifically. When Pets.com[1] went down in early 2000 it wasn't neither the idea, nor the tech stack that brought the company down, it was the speculative business dynamics that caused its collapse. The fact we've swapped technology underneath doesn't mean we're not basically falling into ".com Bubble - Remastered HD Edition".

    I bet a few Pets.com exec were also wondering why people weren't impressed with their website.

    [1]https://en.wikipedia.org/wiki/Pets.com

    • simianwords 2 days ago

      Do you actually want to get into the details on how frequently do markers get things right vs get things wrong? It would make the priors a bit more lucid so we can be on the same page.

classified 2 days ago

Exactly. Microsoft for instance got a noticeable backlash for cramming AI everywhere, and their future plans in that direction.