wondering where that eval is; human data @ OAI; views my own
Joined May 2026
- Tweets68
- Following171
- Followers793
- Likes90
Wyatt retweeted
New voices. Better accents. And less "checking..."
GPT-Live-1.1 brings these updates to Voice in ChatGPT, Codex, and dots, with new voices in French, German, Hindi, Mexican Spanish, Korean, and Japanese.
Rolling out today to all users. Coming soon to the API.
To anyone studying math and worried about the opportunity zone—I hope it’s a comforting anecdote that in my experience I meet very successful math grads in every field/department at every company and every hiring manager I’ve ever met loves math grads. Everything is math—and working with abstract concepts forces a habit of clear communication.
I hope this is common knowledge, but sending good mojo to young mathematicians navigating a ton of change in an ancient field!
Wyatt retweeted
What, the headline doesn’t match what the article measures at all?? If OAI raised Sol’s API costs by 10x, this graph would present OAI’s “subscription value” as becoming 10x higher despite no consumer benefit lmao
Anthropic Subscriptions Offer 5x+ More Value Than OpenAI
Limit testing every AI subscription plan from Anthropic, OpenAI, Meta, SpaceXAI, MiniMax, Moonshot, Zdotai, Cursor, and Cognition
newsletter.semianalysis.com/…
Wyatt retweeted
Millions of people turn to AI to navigate difficult relationships, work through stress, or support someone they care about.
We’re introducing MentalHealthBench, an open benchmark developed with mental health experts to measure how well AI responds in these moments. 🧵
Wyatt retweeted
A common trend is that once we measure something well, the models will saturate the performance on that metric within a year. I really hope this happens for model performance in mental health situations!
Also congrats Luna, model powering billions of users and topping the chart!
We’re demonstrating how frontier models have continued to improve in realistic mental health conversations with MentalHealthBench.
This new open benchmark was built with input from more than 80 mental health clinicians.
We’re releasing it openly so other researchers can examine the methods, run their own evaluations, and build on the work.
openai.com/index/introducing…
Thinking about my late great grandmother, Flo—she was know for being permanently rude after surviving Great Depression+Dust Bowl Combo. A great anecdote from a thanksgiving meal in 1971:
Her grandson Craig had gained weight at college, and he was sensitive about it. The family warned her not to bring up his weight as it would upset him.
As Craig walks into the room, his grandmother, with an acre-wide grin, asks:
“Well goddamned Craig, how fat are ya?”
Something to be said for the rascal grandparent
Connectors are fun— I went to chat complaining my toe hurts expecting sympathy and got “you ran 11 miles yesterday bro what do you expect”
There’s like a woo-to-Q phenomenon with EA, I found EA philosophy a tad impractical (I’m dumb, that’s why I need a compass and a mechanism for some self forgiveness), but now I’m gonna have to go to lighthaven just to unalign with the maniacs
Wyatt retweeted
The AI-assisted intelligence report was “entirely false,” per 1 of the sources.
But it also “almost started a war,” the source said.
Any US operation against a Chinese vessel could have risked spiraling into an armed conflict between the 2 nations.
cnn.com/2026/09/18/politics/…
