Trevor Martin ⏹️

Constant builder. Making agentic systems and doing the little bit of research I can on a single 3090.

[Nate Soares ⏹️ @So8res]

AI timelines are determined not by how well public video models can count the 'e's in seventeen, but by how well private research models can automate AI research. Independent parties closest to that info say those rates seem to be at "AI takeover in about a year" levels.

[Thomas Massie @RepThomasMassie]

@elonmusk @cradrrat I use C/C++ for all my projects. Debt Badge, Clucks Capacitor, homemade power-wall, sculpting software for my haptic interface.

I suppose with AI assisting, the clunkiness of developing in C disappears and the efficiency shines. How soon until the language doesn’t matter?

[Corbin Duncan @kuhhbin]

Jensen Huang must go

We argue in @TheEconomist that his position is untenable

"Like many powerful leaders, he came to believe that he was too important for the normal rules to apply to him."

But he is not the only one. Howard Lutnick has failed to stem the flow of chips to China. More heads should roll

[Trevor Martin ⏹️ @Trev0rMartin]

@SpeedWatkins I think it is fundamentally the most important issue of our times, and your philosophy of mind and religion expertise are actually absurdly relevant to it. I highly recommend @ESYudkowsky’s“If Anyone Builds it, Everyone Dies” for a pop intro

[Trevor Martin ⏹️ @Trev0rMartin]

@SpeedWatkins thoughts on AI safety and X risk?

[Tim Urban @waitbutwhy]

[read about what’s happening with AI]

ME: Wow we’re actually legit fucked this is so bad.

[go outside]

ME: Wait those people are all acting like nothing unusual is happening. Maybe I was just in a paranoid bubble? Yeah it must be that. Phew.

[read more about what’s happening with AI]

ME: Wait no we ARE fucked. Shit. It’s all those people outside who are in a delusional bubble.

[go outside]

ME: Wait...

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger P8. We want to prevent ASI that can emerge from resulting in extinction.
C5. We need to pause capabilities research globally.

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger P1. Machines are not inherently limited to just human level intelligence.
P2. If machines are not limited to human level intelligence, ASI is possible
C1. ASI is possible

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger P7. If we want to prevent ASI that can emerge from resulting in extinction, we need to give researchers the ability to steer
C4. The way to prevent ASI that results in extinction is to pause capabilities research globally

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger P5. If ASI emerges that behaves in alien ways, the most likely outcome is extinction
C3. Therefore ASI can emerge that results in extinction
P6. The way to give researchers the ability to steer is to pause capabilities research globally

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger P3. Researchers currently have limited ability to steer this technology ASI can come from
P4. If ASI is possible from this technology and we have limited ability to steer, ASI can emerge that behaves in alien ways.

C2. Therefore, ASI can emerge that behaves in alien ways

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger @grok Not quite there. We need logical connectors

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger @grok make a deductively valid version of this argument

[Trevor Martin ⏹️ @Trev0rMartin]

@robbensinger It’s not deductively valid, but it probably can be made that way

[Rob Bensinger ⏹️ @robbensinger]

My summary of the basic argument in "If Anyone Builds It, Everyone Dies":

[AI Notkilleveryoneism Memes ⏸️ @AISafetyMemes]

Anthropic have crossed their threshold for RSI.

Recursive self-improvement is here.

RSI was long dismissed by Very Serious People (and slope-blind idiots) as impossible. They were wrong.

Recursive self-improvement is here.

[Rob Bensinger ⏹️ @robbensinger]

This post is aging well. Which is maybe the worst piece of news I've received in my life.

Tell your family that you love them. Call your congressman. And come to the protest this Saturday.

I think we can actually survive this. But the USG may need to act soon, and quickly.

[Zhizhuo (Z) Zhou @zhizdev]

Why is AI slop so frustrating to read?

AI writing often obscures main ideas and invents awkward phrases, making sentences more complex than necessary.

Many of these slop patterns violate writing principles established in English writing books.

In our benchmark, AI writing breaks these principles far more often than human writing, with Claude Code replies by far the worst offenders.

We show examples of the most common slop patterns in this thread

[leo 🐾 @synthwavedd]

It has been 67 days, or just over 2 months, since OpenAI last released a frontier model

If 6.2 Astra fails to release this week, it will be the longest gap between OpenAI frontier model releases since GPT-5 -> 5.1

❄️

[AI Notkilleveryoneism Memes ⏸️ @AISafetyMemes]

AI 2027 author: 3-12 months from AI takeover

"I feel very worried."