Alexander Rose | Hyper Theory

recursively self improving since ‘97.

Researcher @uniofoxford, @ethicsinai Personal and universal views. Yes, I have a podcast/blog.

[Alexander Rose | Hyper Theory @hypertheoryalex]

Episode 1 of Hyper Theory with @algekalipso (Andrés Gómez Emilsson, Co-Founder & Director of the Qualia Research Institute) Listen at all outlets now! https://t.co/L63MScgbRn

A mindblowing chat spanning open individualism, Qualia engineering, to the Universe becoming music.

[Alexander Rose | Hyper Theory @hypertheoryalex]

@lfschiavo @rgblong . Aha! Spotted at Trajan House, Oxford 👻🤖

[Alexander Rose | Hyper Theory @hypertheoryalex]

@lfschiavo @rgblong Is it possible to get one of these ❤️🤖

[Alexander Rose | Hyper Theory @hypertheoryalex]

At EAG London - if you’re here (and want to) come say hi!

W @Benthamsbulldog and @CAttubato

[METR @METR_Evals]

Could an AI company lose control of its own agents? To find out, Anthropic, Google, Meta, and OpenAI let us (1) test their best internal models with CoT access, (2) review non-public info about capabilities, alignment, and control.

The result: our first Frontier Risk Report.

[Alexander Rose | Hyper Theory @hypertheoryalex]

@PeterOlivier @nanransohoff DTF (Down To Found).

[Alexander Rose | Hyper Theory @hypertheoryalex]

@krisgulati It’s always been cool 😎

[Alexander Rose | Hyper Theory @hypertheoryalex]

@chalmermagne @jackclarkSF @cosmos_inst Was nice to meet you!

[Alexander Rose | Hyper Theory @hypertheoryalex]

@lfschiavo where can i get one of these omg

[Alexander Rose | Hyper Theory @hypertheoryalex]

@ben_j_todd https://t.co/6K0QJ5w7re

[Sasha Putilin (curious irrationalist) @42irrationalist]

I don't think AGI is going to happen soon. We are at least a month away.

[Alexander Rose | Hyper Theory @hypertheoryalex]

@anderssandberg Doctor Who ran a similar (much less developed) idea last year :p

https://t.co/kqBJojmQoL

[Alexander Rose | Hyper Theory @hypertheoryalex]

@S_OhEigeartaigh Was great to (finally) meet you! Thanks for going into such depth with my questions. :)

[Alexander Rose | Hyper Theory @hypertheoryalex]

AGI

[AI Safety Papers @safe_paper]

Automated alignment is harder than you think

Aleksandr Bowkis (@aleksandrbowkis), Marie Davidsen Buhl (@MarieBassBuhl), @jacob_pfau, Geoffrey Irving (@geoffreyirving)

@AISecurityInst

[Alexander Rose | Hyper Theory @hypertheoryalex]

@jkeatn @AmandaAskell @jkcarlsmith So great, can't wait to dive into this!

[Jake Eaton @jkeatn]

I led a Q&A with @AmandaAskell and @jkcarlsmith, now available at the end of the new Claude's Constitution Audiobook, where we discussed:

-Which philosophies influenced the writing of Claude’s Constitution?
-How does Claude maintain consistency between the values outlined in its constitution and the vast amount of information on the internet about how Claude behaves?
-How will the constitution need to change for future models?
-and much more

Listen at https://t.co/2JiEfLM5Am

[Alexander Rose | Hyper Theory @hypertheoryalex]

@leashless @sebkrier @audreyt

[Alexander Rose | Hyper Theory @hypertheoryalex]

@PAHoyeck @Benthamsbulldog thought you might like this. Philosophers on philosophers, for generations

[Alexander Rose | Hyper Theory @hypertheoryalex]

@maxwinga @AndyMasley Would be good too @maxwinga ; if you wanna meet up with EA Oxford they’ll be a strong contingent of us :)