Jump in the discussion.

No email address required.

o1 is the saddest :marseyitsover:thing. They hyped up this new model for ages as some mysterious "next step towards AGI", and it turns out to be yet another chain-of-thought prompt with the big innovations being that

  • They hide the reasoning tokens from you

  • They benchmark it against other models that don't use chain-of-thought

  • They still charge you for the uncontrollable, invisible reasoning tokens :marseymerchant: and apparently they freak out if you try to see them lmao

:soyjakwow: WAOW Sam's done it again AGI is just around the corner!!

As for actual intelligence, every benchmark is suspect at this point, I've seen too many meme models shoot ahead on them while being completely r-slurred. Even the chatbot arena is being gamed by now. The only metric I trust is to look at whatever model the chatbot gooners :marseycoomer2: are swarming around because if nothing else, convoluted degenerate fetish ERP requires actual intelligence and vendors aren't trying to game it. I see no buzz about o1 there so it's a nothingburger, Anthropic is mogging Sam, you heard it here first

Jump in the discussion.

No email address required.

The only metric I trust is to look at whatever model the chatbot gooners are swarming around because if nothing else, convoluted degenerate fetish ERP requires actual intelligence and vendors aren't trying to game it.

Free market mechanisms win again :#marseywholesome:

Jump in the discussion.

No email address required.

>As for actual intelligence, every benchmark is suspect at this point

More than that, benchmarks are borderline completely useless in ai atm. Grifters have finally figured out how to wrangle the models so they look good on a chart with abstracted nerd metrics that no one can replicate on their own machines.

MY 128B MODEL IS 23.4% BETTER THAN CLAUDE 3.5 IN GSM8K, IFEVAL, AND IT CAN SPELL STRAWBERRY CORRECTLY. GIVE ME TRILLION IN FUNDING NOW. :soyjakyell:

Ai will be exclusively filled with scams and business majors for months/years before any real products that make profit are made. If you havent seen it, check out the "Reflection" model that came out a few days ago. CEO straight up just lied about the models capabilities. Most likely to scam some money out of vcs.

Jump in the discussion.

No email address required.

Most likely to scam some money out of vcs.

I mean, VCs are still investing in crypto bullshit so they deserve to lose their money

Jump in the discussion.

No email address required.

The only real product I see for AI is auditing models being developed by the big corpo banks rn. !codecels fintech bros will make Anime real just to make them their bangmaid secretaries.

https://kpmg.com/xx/en/our-insights/ai-and-technology/ai-in-financial-reporting-and-audit.html

Jump in the discussion.

No email address required.

I use chatbots for basically the same purpose, to give me some interesting signal to follow up on out of a bunch of noise

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.