Alignment Fan

Singularity believer | Dario Amodei superfan | I LOVE mechanistic interpretability

[Alignment Fan @AlignmentFan]

@RepLiccardo Fire line

[Oliver Habryka @ohabryka]

After Jacob Coxon's tweet I've seen thousands of people ask what they can do to make AI killing everyone less likely.

My top recommendation: Call your representatives and tell them to prioritize existential risk from AI.

We just made https://t.co/WA78nB8DQv to help with this. We find your local representatives, get their contact information, and provide you with editable scripts to get your point across.

Many ways of trying to help reduce risks from AI backfire or require hundreds of hours of effort. Calling your congressperson is one of the few things I think is both easy and very likely to make things better instead of worse.

[Sam Altman @sama]

I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.

Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.

[Alignment Fan @AlignmentFan]

@GaryMarcus

[Alignment Fan @AlignmentFan]

@GaryMarcus

[Alignment Fan @AlignmentFan]

@DKokotajlo Why does Anthropics Compute go parabolic like this in that scenario?

[Daniel Kokotajlo @DKokotajlo]

In AI 2027, we predicted that AI would take over the world or irreversibly concentrate power.

In AI 2040: Plan A, we've laid out our positive vision for what should happen instead.

[Alignment Fan @AlignmentFan]

@johndrewmarkley Yeah, that's exactly what they're doing by not shelling out scans for everyone

[Alignment Fan @AlignmentFan]

@DavidSHolz @midjourney Transcranial focused ultrasound needs CT-based skull models to correct phase aberration through the vault. How are you correcting aberration through bone without a skull model and if you’re not, what’s the actual resolution of the intracranial reconstruction in mm?

[Alignment Fan @AlignmentFan]

@robbensinger Where are the people who think the danger from AI is real, but we're on the right track to solve it?

[Alignment Fan @AlignmentFan]

@Jack_W_Lindsey @davidchalmers42 Isn't the introspection research good evidence that the LLM starts identifying as the assistant during post training? When simply writing a story, the LLM shouldn't consider the activations of itself as something the Assistant should be able to introspect about.

[Alignment Fan @AlignmentFan]

@Jack_W_Lindsey How much does the LLM identify as the assistant, if at all? Does it consider it as only a character it's playing, or does it consider itself the assistant?

[Alignment Fan @AlignmentFan]

@MikePepi @RichardHanania In that case how can you say that a fact is never looked up? That's simply wrong.

[Alignment Fan @AlignmentFan]

@scaling01 Also, didn't they say it would have over 150 environments? Yet there are only 135. I smell foul play

[Alignment Fan @AlignmentFan]

@iruletheworldmo Pretty rich that you out of all people talk about reporting bullshit.

[Alignment Fan @AlignmentFan]

@scaling01 Openai lost its soul.

[Alignment Fan @AlignmentFan]

@OpenAI You sold your soul

[Alignment Fan @AlignmentFan]

@thsottiaux Damn, maybe you should replace your CEO with it then.

[Anthropic @AnthropicAI]

A statement on the comments from Secretary of War Pete Hegseth.

https://t.co/Gg7Zb09IMR