Miles Brundage
@Miles_Brundage
AI policy researcher, @lfschiavo wife guy, Executive Director of @AVERIorg, Substacker, fan of animals and sci-fi, views my own
13.8K Following 78.8K Followers
You met me at a very frontier AI auditing time in my life
It's clear that a lot of people at the AI companies have done some serious reflection in the past few months
And it's also clear that a lot of people have done zero reflection + still think they're making a fun, safe thing that should be shipped as quickly as possible
Show more
new sticker just dropped. find me at the swarm dynamics hackathon if u want one
Who called it losing chain of thought monitorability through opaque internal reasoning and not “weight, weight, don’t tell me”
The task force has been named the “Super Intelligence Force,” Trump’s preferred name for AI, and will have 120 days to prepare the report, the report said
burning tokens
.@dgrobinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specific rules or new laws. We need to talk about culture,” he writes:
Show more
Now Connie Chan is also attacking Scott Wiener for supposedly being weak on AI regulation. You could only think that if you wilfully ignored everything he has done on the subject. I wrote about this earlier this year.
Show more
personal ai for reservations, call that supper intelligence
sam and dario are like my divorced parents who just take turns saying things i like to hear in order to curry favor so i don't embarrass them in family court in front of a judge
I’m afraid that speed will make many lessons from history moot in the case of AI. This is why I think references to Industrial Revolution are useful but also very incomplete. And why my confidence intervals so big.
Show more
Also, as usual it bears repeating that this is not just an OpenAI issue.
If you added up the very best isolated practices for security, safety culture, etc. from each company and combined them in one company, it's still be sub-Chernobyl level IMO
Show more
btw, this is quite huge.
there is no sandbox from any provider today that i would treat as unescapable by a sufficiently capable adversarial agent.
containers fell.
kvm fell.
sandboxing is hard!!!
.@dgrobinson’s essay on why he resigned from OpenAI is very good reading + advances the argument that Silicon Valley’s broader culture is built on panicked sprints & trial and error, which don’t foster the wisdom you need to build advanced AI.
Show more
This is super cool. I'm looking forward to seeing what comes out of it!
They’re calling it the coolest hackathon of the year and it starts in under 2 hrs
AI Village x Grove Research: AI Swarm Dynamics Hackathon!
After the Hugging Face and German Wiki incidents, we need better tools to understand AI swarms. Spend a weekend building them.
$3,000 in prizes
Free compute
October 3-4 🧵
Show more
.@dgrobinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specific rules or new laws. We need to talk about culture,” he writes:
Show more
This was my reaction to that OAI security post also.
No one (informed) denies there are good security people there, the concern that informed people have about OAI and other companies right now is about culture and decision-making processes
Show more
Ah well. Nevertheless
