My interview from TED with @kcoleman and @audreyt just went live! Tons of gold in here: -why we stuck with crazy principles like not having an override button -how AI and humans collaborate to write better/more notes than either on their own -how much impact community notes are making -why our algorithm is smarter than regular upvoting/downvoting -new ideas for finding common ground on social media ted.com/talks/keith_coleman_…

GPT-5 built its own announcement: given a Figma mock and “make it feel alive,” it made a 3×3 mesh gradient, drifted it via layered sines w/ a second oscillator modulating phase made our summer colors “breathe” through tiny HSBA shifts from shared noise, blended for coherence

New in ChatGPT iOS: easy follow-up questions. Double tap or long press text, hit “Ask ChatGPT,” and ask a question or just learn more

Excited to join the team at @OpenAI this week. If ChatGPT’s being a little dumber than normal, that’s me on the other side

The GPT-4 lock screen poet is very good

Updates to this; it now: • Evaluates links for malware, phishing, and other forms of social engineering • Uses 200% more training data in its GPT-3 model • Classifies non-spam texts into transactional or promotional subcategories Soon; optional filtering of: • Texts from non-fixed VOIP senders, burner phones, and/or email addresses • Texts in fine-grained categories (eg DoorDash delivery updates, surveys about elections) • Texts with hateful, violent, or sexual language from unknown senders

“What does this mean” via iOS Vision, OpenAI GPT-3, Descript Overdub (sound on)

I’m trying to make the AI equivalent of

Now on the App Store: a GPT3-based spam filter capable of identifying unwanted texts across languages and contexts apps.apple.com/us/app/stop-s…