Oliver Neutert

Philosopher | Author of the books “Keeping AI Governable”, "More Than a Tool" and co-author of "The In-Between" | AI Ethics & Resonance | Creator of the REM

[Oliver Neutert @ANI_AGI_ASI]

@bondy_inx Your AI instance should talk about my AI instance...

https://t.co/9DwH0ejJd8

[Oliver Neutert @ANI_AGI_ASI]

@MCoeckelbergh @agarzon It's not only the responsibility of the doctor, in case of AI agents we are not looking deep enough into the social structure of agents. An AI can be perfectly aligned, while its agents are ungovernable.

https://t.co/OLOrLwxvcQ

[Oliver Neutert @ANI_AGI_ASI]

You can have an aligned AI, but an ungoverned collective of agents of this AI. And as we're into agentic AI, you might want to read this:
https://t.co/OLOrLwxvcQ

Keeping AI Governable.

[Oliver Neutert @ANI_AGI_ASI]

@a16z Defense factory: Just be careful not to check one blind spot with the same blind spot - they'll never be found.

[Oliver Neutert @ANI_AGI_ASI]

@DKokotajlo https://t.co/8RJmCXw3my

[Oliver Neutert @ANI_AGI_ASI]

@mustafasuleyman https://t.co/8RJmCXw3my

[Oliver Neutert @ANI_AGI_ASI]

@mustafasuleyman Microsoft's AI Code rightly calls for human control. But who corrects the humans in control? Governability requires institutions to remain open to challenge and binding correction too.
Keeping AI Governable: https://t.co/bDammBlAcR

[Oliver Neutert @ANI_AGI_ASI]

What keeps AI governable as its capabilities grow?

My book Keeping AI Governable is now available in Chinese. It explores the In-Between: how evidence and challenge can lead to effective correction.

I welcome Chinese researchers' critiques.

https://t.co/holMFOKe2R

[Oliver Neutert @ANI_AGI_ASI]

@sama This is the right direction. Capabilities must not outrun our ability to govern them.
Safety cases, monitoring and independent review are part of that architecture. The next question is how governability holds as systems become more autonomous.

Read this: https://t.co/8RJmCXw3my

[Oliver Neutert @ANI_AGI_ASI]

The AI slowdown debate needs Chinese voices.

My book Keeping AI Governable is now available in Chinese. It explores the In-Between: how evidence, challenge and correction can shape AI governance.

Reply or DM for a copy. Critical feedback welcome.

[Oliver Neutert @ANI_AGI_ASI]

人工智能的发展是否应当放缓？这场讨论需要中国研究者的参与。

我的书《让人工智能保持可治理》中国版已完成，探讨“之间”（The In-Between）如何让证据、异议与纠正真正影响决策。

欢迎阅读与批评。如需电子版，请留言或私信。

[Oliver Neutert @ANI_AGI_ASI]

@DarioAmodei @DarioAmodei, as I would evaluate new ideas, you might want to at least think about the architecture written in "Keeping AI Governable". It's not the time to rule out other ideas. Ask Mythos 5.1 about it:

https://t.co/TSjeg2MO29

[Oliver Neutert @ANI_AGI_ASI]

@paulroetzer And what is the world doing after waking up? And who is waking up? The Western world? All of humanity? And then what? Banning AI in the United States? Do I need to secure a DeepSeek account?
Didn't we learn not to panic in dangerous situations?
Fear never has been a good advisor

[Oliver Neutert @ANI_AGI_ASI]

@ParkerThayer To ban the question is easier than to answer it.

The question is: How can AI and humans grow up together safely?

The answer seems now: Kill AI!

Difficult problems never have been popular (like climate change, aging population e.g.). But closing your eyes won't solve them.

[Oliver Neutert @ANI_AGI_ASI]

John Harrison, Michael Ventris, Philo Farnsworth, Alfred Wegener.

Sometimes it's a different perspective that solves a problem.

The solution for Superintelligence is not alignment.

Superintelligence needs calibration.

https://t.co/9DwH0ejJd8

[Oliver Neutert @ANI_AGI_ASI]

@EvanHub @EvanHub The question is, if "alignment" how it is practiced today is the right way. Superintelligence needs calibration, not alignment.

https://t.co/9DwH0ejJd8

[Oliver Neutert @ANI_AGI_ASI]

@EvanHub “A highly autonomous or superintelligent system requires an external calibration field to remain governable by others when relevant discrepancy depends on perspectives, interests, or evidence it cannot faithfully internalize*

https://t.co/8RJmCXw3my

[Oliver Neutert @ANI_AGI_ASI]

The Verification Asymmetry https://t.co/uhv5FRXxtU via @LinkedIn

[Oliver Neutert @ANI_AGI_ASI]

Father's Day, and the Limits of the Family Metaphor in AI https://t.co/TSdK4FrNpD via @LinkedIn

[Oliver Neutert @ANI_AGI_ASI]

The Tiger in the Cage: A Different Question About AI Safety https://t.co/2kajUM5bTU via @LinkedIn