Rohan Viswanathan

We’re updating our ChatGPT Pro and Plus subscriptions to better support the growing use of Codex. We’re introducing a new $100/month Pro tier. This new tier offers 5x more Codex usage than Plus and is best for longer, high-effort Codex sessions. In ChatGPT, this new Pro tier still offers access to all Pro features, including the exclusive Pro model and unlimited access to Instant and Thinking models. To celebrate the launch, we’re increasing Codex usage for a limited time through May 31st so that Pro $100 subscribers get up to 10x usage of ChatGPT Plus on Codex to build your most ambitious ideas.

north-of-San-Francisco style

Arguably the greatest poster in film history.

Has anyone else noticed that ChatGPT/Open AI is quietly building a better Wikipedia?

Paul Cornoyer, Christmas in Madison Square Park

state-aware product cards 💡 slept on posting my take w/ view transitions 🧑‍🍳

@tszzl @aidan_mclau Big hero 6 was onto something

@aidan_mclau it’s just anywhere where you can get more on policy data will strictly dominate, so highly disposable bots, bots that have very little sim2real gap should be where the density of experiments happen

there is probably a bitter lesson of robotics.. that composable, easy-to-scale things like nanobots are way better than trying to build a humanoid that can do backflips

@derekzxu This is awesome!

check out my first blog post here about a simple Eval Protocol workflow for llm-as-judge. turn “this seems off” into a reproducible test and a reliable metric: fireworks.ai/blog/ai-benchm…

@CadeGordonML @Anthropic Love u Caden excited to hang in SF

Excited to share that I'll be joining @Anthropic to work on pretraining science! I've chosen to defer my Stanford PhD, where I'm honored to be supported by the Hertz Fellowship. There's something special about the science, this place, and these people. Looking forward to joining some of my most brilliant and compassionate colleagues!

@thomas_liang_ What does this mean

@pgasawa ong

I wonder if MAA problem writers know that the 30 problems they wrote for this week and the next are not just for USA(J)MO qualification but will benchmark AI progress for the next year (“AIME25”) 🙃

Excited to share our work on calibrated protein search and genome mining with conformal prediction, now in @NatureComms! 📎 rdcu.be/d5pJG 💻 github.com/ronboger/confo… (1/3)

@thomas_liang_ Real

@Kashyap4567 @ashleyyjoelle Yea the people watching was going crazy but I’m bored now

@ashleyyjoelle @rpi2b this sounds like your situation lol

Anyone know what to do alone in SFO… just deplaned my red eye to the east coast 😭

Check out this course we developed @BerkeleyML with @udacity!

ML@B is super excited to have partnered with Udacity on their Generative AI Nanodegree Program which launched today! It covers various topics, enabling developers to integrate generative AI into software applications. Check it out here: tinyurl.com/yc63zb88!

ML@B is nothing without its people

@BerkeleyML is super excited to be hosting our annual Career Fair this September! In the past, we’ve enjoyed collaborating with companies like @nvidia, @Citadel, and @shieldaitech. If you're a company interested in participating, dm us!

@rovirmani i went in ny lol that’s the most I’ve laughed in a while

yo who tryna go to the south park musical

@percyliang Google tried a version of this with Duplex (ai.googleblog.com/2018/05/duplex…) but they decommissioned it for now at the end of last year. At its inception there was concern about the ethics behind the second party not knowing they were talking to an AI in a “natural” conversation.

ChatGPT is reactive: user says X, ChatGPT responds with Y. Risks exist but are bounded. Soon it will be tempting to have proactive systems - an assistant that will answer emails for you, take actions on your behalf, etc. Risks will then be much higher.