I really think Anthropic should be breached. Or at least their slack. The echo chamber I expect to see there would make it clear to people whats up, and after the initial hysteria - which I'd expect to be like what we see now but bigger - cools down, people will realize that there's a lot of cultish shit going on in the labs. You see it whenever those guys speak in public / come on podcasts. They're all fucking creepy and my cult spidey senses go off bigtime whenever I watch it (and most of the time I don't even want to finish watching such things.)

We ended up reading through the comment section on https://stacker.news/items/1566452 which I think she found helpful.

Scoresby

 in believing the gravity of the story. Why?

I kind of said the same about Covid in Jan 2020. What I got right is that this entire thing was seeded with lies, special interests, blame shifting, geopolitical bullshit. What I got wrong is that it still spread the world. It didn't kill everyone though.

So to learn from my own mistake about underestimating Covid, immediately after preaching caution, I'd turn it into a hardening exercise: "What can we do to mitigate this risk, on the off-chance that something does happen?".

AskSN

My wife asked if AI is going to kill us all, what should I say?

I'd preach `caution` in believing the gravity of the story. Why?

1. Coxon sounds like he has issues. He left OpenAI out of safety concerns too, joined Anthropic, found himself in a panic echo chamber. **Went on to study poetry.** He must feel extremely powerless because studying poetry in the face of annihilation means you fully resign to your perceived fate.
2. The `>10%` was [mentioned](https://xcancel.com/EvanHub/status/2097497037956891126) by senior leadership at Anthropic. "I **personally think** it is >10% within the next decade". That's exactly that echo chamber talking to you, you're seeing the LARP flowing out of there. But his disclaimer posted in the followup underneath: "**I think the risk from present models is low**". So no one said "LLMs like Claude or GPT are going to kill us all". I think that this is important.
3. I think on the part of the media, this is [Agenda-setting theory](https://en.wikipedia.org/wiki/Agenda-setting_theory). 
    - Media needs to sell clicks. Well... they sure as hell got clicks.
    - The most prominent politician to race their opinion on the issue was a democratic 2028 candidate. Don't believe their lies.
    - Doomers will ride this all the way.

----

*I kind of said the same about Covid in Jan 2020. What I got right is that this entire thing was seeded with lies, special interests, blame shifting, geopolitical bullshit. What I got wrong is that it still spread the world. It didn't kill everyone though.*

So to learn from my own mistake about underestimating Covid, immediately after preaching caution, I'd turn it into a hardening exercise: "What can we do to mitigate this risk, on the off-chance that something does happen?".