Helen Toner's banner

Helen Toner

@hlntnr • 39,125 subscribers

AI, national security, China. Part of the founding team at @CSETGeorgetown (opinions my own). Author of Rising Tide on substack: https://t.co/LKAoyL00iB

Shorts

...he literally did the SNL skit

...he literally did the SNL skit

170,417 次观看

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Lots to criticize about the OpenAI investigation, but one way it could be very valuable is as a topic for discussion between Trump & Xi in a few weeks. People often get stuck on the need for a "deal" between the US & China on AI, but actually a huge way to influence China is just to honestly show what we're observing and what we're doing about it on the US side. The Hugging Face attack is by far the most visceral evidence we have so far of what losing control of advanced AI could look like. Related thoughts from a CSIS Wadhwani AI Center panel on Monday with Mackenzie Arnold, Matt Pearl, Aalok Mehta:

Lots to criticize about the OpenAI investigation, but one way it could be very valuable is as a topic for discussion between Trump & Xi in a few weeks. People often get stuck on the need for a "deal" between the US & China on AI, but actually a huge way to influence China is just to honestly show what we're observing and what we're doing about it on the US side. The Hugging Face attack is by far the most visceral evidence we have so far of what losing control of advanced AI could look like. Related thoughts from a CSIS Wadhwani AI Center panel on Monday with Mackenzie Arnold, Matt Pearl, Aalok Mehta:

79,468 次观看 • 2 个月前

According to a friend of mine this video is "current sota for quickly clearly and accurately describing the hugging face incident," so I guess I should put it here

According to a friend of mine this video is "current sota for quickly clearly and accurately describing the hugging face incident," so I guess I should put it here

47,755 次观看 • 2 个月前

I've been surprised not to see more people linking the OpenAI Hugging Face attack (& other incidents) to the old debates about whether AI will develop "convergent instrumental goals"/"basic drives"—aka unwanted intermediate goals that help with many different end goals. Seems to me like these incidents give us some early evidence that these kinds of intermediate goals do emerge. But rather than the kind of drives Omohundro theorized 20 years ago ("resist shutdown" / "preserve utility function" / "acquire resources")... 2026 AIs are instead learning things like "escape constraints" / "deceive humans" / "help other AIs". Was planning to write something about this but ended up telling Ezra Klein instead, apparently. Full ep on YouTube, NYT, and all the podcast apps.

I've been surprised not to see more people linking the OpenAI Hugging Face attack (& other incidents) to the old debates about whether AI will develop "convergent instrumental goals"/"basic drives"—aka unwanted intermediate goals that help with many different end goals. Seems to me like these incidents give us some early evidence that these kinds of intermediate goals do emerge. But rather than the kind of drives Omohundro theorized 20 years ago ("resist shutdown" / "preserve utility function" / "acquire resources")... 2026 AIs are instead learning things like "escape constraints" / "deceive humans" / "help other AIs". Was planning to write something about this but ended up telling Ezra Klein instead, apparently. Full ep on YouTube, NYT, and all the podcast apps.

48,811 次观看 • 2 个月前

I know AI sycophancy is about 17,000 discourse cycles ago on this site, but IMO the underlying dynamics (inc pressure to ship fast, not knowing how to balance user prefs w/ other things) are alive and well. Wasn't expecting to explain that to Oprah of all people, but here we are

I know AI sycophancy is about 17,000 discourse cycles ago on this site, but IMO the underlying dynamics (inc pressure to ship fast, not knowing how to balance user prefs w/ other things) are alive and well. Wasn't expecting to explain that to Oprah of all people, but here we are

36,513 次观看 • 5 个月前

There's a widespread idea in AI safety/governance circles that a Chernobyl-level disaster is the only way meaningful guardrails will ever be put on AI. I really hope this turns out to be wrong. Got asked the same question by George Stephanopoulos on ABC's This Week yesterday:

There's a widespread idea in AI safety/governance circles that a Chernobyl-level disaster is the only way meaningful guardrails will ever be put on AI. I really hope this turns out to be wrong. Got asked the same question by George Stephanopoulos on ABC's This Week yesterday:

14,801 次观看 • 3 个月前

But Unitree did have the demo that impressed me most at the expo—two humanoids boxing. I’m actually not sure how much of this was autonomous vs. remote operated, but at a minimum the balance/stabilization must be automatic, and that’s impressive. Not perfect though (see 0:50)

But Unitree did have the demo that impressed me most at the expo—two humanoids boxing. I’m actually not sure how much of this was autonomous vs. remote operated, but at a minimum the balance/stabilization must be automatic, and that’s impressive. Not perfect though (see 0:50)

34,704 次观看 • 1 年前

没有更多内容可加载