Mysterious stranger, and product nerd. formerly Mozilla; DocumentCloud
Oakland
Joined March 2008
- Tweets73.9K
- Following5K
- Followers4.4K
- Likes94.2K
Ted Han ★ 韓聖安 retweeted
The right approach for AI regulation should start with something like the NTSB
Haize began as an independent safety evaluator for frontier models, so I’m certainly sympathetic to Dario’s Pace the Frontier proposal
but I’m skeptical embedded evaluators are sufficient
some questions and four possible alternatives:
leonardtang.me/blog/alternat…
Ted Han ★ 韓聖安 retweeted
Dear friends who are using one of the AI design systems: it is very easy to pay a few bucks for a typeface from a service like @futurefonts. You’ll look a little different and make the world a better place!
I'm linking my keytrace.dev: did:plc:l5yz32nydpebjlcdfgycmf3x
Ted Han ★ 韓聖安 retweeted
1. Roll out articles so people can write really long tweets. These will mostly be written by AI
2. Then, use AI to summarize those articles back into tweets
A 100M$ dollar raise that somehow includes both Bain Crypto and The Knight Foundation.
The future is amazing; much stranger than i imagined.
Last April, Bluesky raised $100M in Series B funding led by Bain Capital Crypto. Since then: 13M → 43M+ users. Join us. bsky.social/about/blog/03-19…
Ted Han ★ 韓聖安 retweeted
I think Sinofsky’s insight extends to social artifacts in general. Once everyone can do something, you aren’t differentiated. Once you aren’t differentiated, nobody cares...
This quoted post is unavailable.
I am much more vehement about this. I think that OpenAI's product choices are deeply deeply antisocial and inhumane.
The way that they've implemented chat bots are isolating, sycophantic and have no reality in your relationships with others save for how you represent them.
These are _choices_ that OpenAI, Anthropic et al have made. There's no reason these technologies _have to_ work like this.
Even Grok here is at least mildly more social in terms of how people interact with it.
The ATProto dev community is on the cusp of some amazing new things.
Dan Abramov imploded his account here, but he just published a deep dive into the promise of ATProto: overreacted.io/open-social/
Ted Han ★ 韓聖安 retweeted
This incredible incompetence of the FBI under Kash Patel seems like a huge story.
Crazy that this guy was literally talking about his plans to shoot up this church for a month on the open internet, but @FBIDirectorKash and @PamBondi are too busy going after imaginary MS-13 Home Depot workers, if only they were doing their jobs instead. :(
A new paper and survey if scientist by @WhySharksMatter and Dr Julia Wester
southernfriedscience.com/twi…
Ted Han ★ 韓聖安 retweeted
Breakneck is full of original reporting, smart insights, and details about China I can’t find anywhere else. @Noahpinion and I are hosting a chat on 8/22 with @danwwang before the book comes out. Details here: lu.ma/gk9j5v06
