Ethan Shaotran (@EShaotran) / X

AI now responds to 50% of my emails. 🤯 Crazy exchange on @gmail between Spark and a colleague to coordinate a @googlecalendar mtg, set the @Zoom, understand timezones. @shadcn for UI, @vercel for AI, @ido_pesok @rauchg @svpino @t3dotgg @jerryjliu0 @jaredpalmer

We're introducing Aligned, a governance platform for AGI, via @OpenAI 's Democratic Grant. After months of hush hush, OAI meetings, and intense building, we're sharing our research findings for aligning AI on a global scale. 1. 93% consensus across 700 global users. 2. 9% of

OpenAI

@OpenAI

Jan 16, 2024

We provided grants to 10 teams who developed innovative prototypes for using democratic input to help define AI system behavior. Read about our learnings and how we'll put this work into practice. openai.com/blog/democrati…

The Mac is back! Quick little weekend project to bring MacOS to life with GPT4. Give it any command and watch your computer interact live. 🤩 You can build this in 2 hours! Or run it in 2 minutes (open-source code below). Inspired by @karpathy.

00:00

Great to shake hands with @JustinTrudeau at #MKT4Intel Machine Learning Conference @CanadianPM @creativedlab @rotmanschool #Trudeau

Ethan Shaotran

@EShaotran

Oct 17, 2024

@elonmusk can we get this feature on X? 😙

xAI

@xai

Oct 17, 2024

Team eXpect was the runner-up at our hackathon, earning $5,000 in API credits. Congratulations! @raydelvecc @raghavp @EShaotran @GiovanniMDanto2

00:00

Harvard's @thefuturesoc and I released a case study on #AI in #WallStreet, which was just published as part of @cccalum's book "Stories from 2045". Check it out: storiesfrom2045.com

Ethan Shaotran

@EShaotran

Jan 17, 2024

Replying to @EShaotran

An overdue shoutout to collaborators and friends @woj_zaremba @ThankYourNiceAI Teddy Lee @jaybaxter @metaviv @colinmegill @Werdnamai @bakkermichiel (and of course Energize's @ido_pesok, Sam, Emi). Thanks for betting on the team of college students!

Replying to @karpathy

The Foundation series!

Replying to @cracked_eng

Looking forward to it!

How will humanity control AGI? Alignment asks how to steer ML models. But there's an important caveat: "who gets to drive?" Our 10 teams have been working on a solution with @OpenAI's grant. As requested, a summary:

OpenAI

@OpenAI

Jan 16, 2024

We provided grants to 10 teams who developed innovative prototypes for using democratic input to help define AI system behavior. Read about our learnings and how we'll put this work into practice. openai.com/blog/democrati…

Excited to announce that after 2 years, our paper “Gesture Learning for Self-Driving Cars” was published! Thanks to my co-authors Jon Cruz and VJ Reddi, the folks at @IEEEorg, and @Harvard SEAS. Read the full paper at ieeexplore.ieee.org/abstract/docum…. #AI #research #selfdriving #harvard

Ethan Shaotran

@EShaotran

Jan 17, 2024

Replying to @EShaotran

Among many takeaways, we found only 9% of proposed guidelines were accepted. This makes consensus is even more important! We believe these AI-enabled approaches to governance have profound implications for not only @OpenAI, but also participatory democracy itself. More to come.

Replying to @EShaotran

The result varied across teams: case repositories for case law, guidelines, voting results, etc. All are important. These findings then get implemented via RLHF, CAI, and eventually harder ML alignment techniques. Anthropic's post (anthropic.com/news/collectiv…) walks through this.

Replying to @rbhar90

You should try energize.ai then - an agent for email scheduling.

85