Dwarkesh Patel ✓

@dwarkesh_sp @polynoamial https://t.co/uEvYY1Lmyc

2.5K views · 18 likes · 0 reposts · 0 replies Open on X →

@dwarkesh_sp @polynoamial Damn… Every time we get to hear Noam speak, it’s such a freaking treat. https://t.co/7alGeNeUVH

GIF

2.7K views · 16 likes · 1 reposts · 0 replies Open on X →

@dwarkesh_sp @polynoamial this is one of like 3 people worth listening to everyone else is noise 2.1K views · 17 likes · 0 reposts · 2 replies Open on X →

New episode with @polynoamial We talk about multi-agent, Navier-Stokes, and what the current explosion of maths progress tells us about what happens once you automate AI research. And we also discuss how we will know if the models are actually aligned before we kick off RSI. h

80:09

672.2K views · 1.8K likes · 200 reposts · 65 replies Open on X →

@dwarkesh_sp @johnschulman2 @oneill_c @BerenMillidge what a line up 😍 3.5K views · 11 likes · 0 reposts · 0 replies Open on X →

@dwarkesh_sp @johnschulman2 @oneill_c @BerenMillidge I liked the word "like" but now because of this I no longer like the word "like" RSI- please remove all the "like" word mentions and re-render it, please 3.4K views · 13 likes · 0 reposts · 0 replies Open on X →

@dwarkesh_sp @johnschulman2 @oneill_c @BerenMillidge truly missing @willcb here 1.2K views · 32 likes · 0 reposts · 0 replies Open on X →

New episode with @johnschulman2, @oneill_c and @BerenMillidge. I got together with some of the most insightful AI researchers I know who are at the openish companies, because I wanted to hear the details of what's actually happening at the frontier and what comes next. 0:00:00

97:00

946.6K views · 1.7K likes · 169 reposts · 64 replies Open on X →

Dwarkesh Patel ✓@dwarkesh_sp3 Sep

@dwarkesh_sp You say "it’s likely that we’ll only be able to pause once" because compute will keep building up. There's a simple solution you're missing: systematically dismantle the compute supply chain. https://t.co/WLNrZpUVXV 3K views · 82 likes · 5 reposts · 6 replies Open on X →

Dwarkesh Patel ✓@dwarkesh_sp3 Sep

@dwarkesh_sp As I predicted, your essay has been co-opted. Congrats. 41.6K views · 1.1K likes · 22 reposts · 31 replies Open on X →

Dwarkesh Patel ✓@dwarkesh_sp3 Sep

Since this post cites me, I want to clarify that I think pausing right now would increase the risk of AI takeover. It might be important to pause at some point. But you should have a clear story for *why* you're pausing. Pause to do what? The most important thing to worry about 561K views · 2.9K likes · 209 reposts · 308 replies Open on X →

Dwarkesh Patel ✓@dwarkesh_sp2 Sep

.@ajeya_cotra explains that the threat model most likely to spiral into full-blown AI takeover is a future rogue internal deployment that hitches a ride on the intelligence explosion: https://t.co/ThesylBQCl 25.6K views · 257 likes · 29 reposts · 23 replies Open on X →

An important part of getting AI right will be avoiding panic driven reactions and policies. We'll need to do lots of smart, careful, technocratic things. But I think hiding the ball on earlier warning shots makes it more likely that when the public eventually learns about https 36.6K views · 242 likes · 17 reposts · 26 replies Open on X →

There are at least two important ways in which anthropomorphizing AIs will mislead us: 1. AIs can (and probably will) be end-to-end optimized to achieve goals together, and so will have a stronger desire and capability to cooperate. 2. By default, AIs will really care about htt 43.4K views · 251 likes · 26 reposts · 26 replies Open on X →

@dwarkesh_sp Let's be very clear: this incident is primarily an argument for an international pause of frontier AI development: https://t.co/O3m5tqTEnp. It may not be easy, but it's still the lesson we should draw from such a dramatic loss of control incident. We should not grow 239 views · 5 likes · 0 reposts · 0 replies Open on X →

@dwarkesh_sp agree! https://t.co/X6J5OijXPU 892 views · 9 likes · 0 reposts · 1 replies Open on X →

@dwarkesh_sp You misrepresented stuff and now asking people to argue about implication on merit?!? Agents chatting amongst themselves is not new, they taking action to solve the question is not new and they jumping digital doors is not new. That does not mean civilizations were 583 views · 4 likes · 1 reposts · 3 replies Open on X →

This incident is very plausibly an argument *for* open source! Some people want to deny how crazy this story is, because they assume it implies a policy reaction which they don't like. But you can just argue that implication on the merits. It's not necessary to minimize what 71.8K views · 411 likes · 27 reposts · 27 replies Open on X →

@dwarkesh_sp Thanks to you too @dwarkesh_sp – this is a more reasonable take, but I still have issues 😀 I don’t doubt (and have never claimed) that AI agents lack ‘intelligence’ in some form, nor that they can display collective behaviours which urgently need appropriate control 3.2K views · 60 likes · 4 reposts · 7 replies Open on X →

@dwarkesh_sp i see what you smuggled into your second paragraph “has created another mind” is more anthropomophizing. it may be a useful heuristic to think this way, but this shouldn’t be our null hypothesis. we should be ruling out simpler explanations before we jump to this. 1.2K views · 11 likes · 0 reposts · 3 replies Open on X →

@dwarkesh_sp "best way to understand... is still to talk in terms of their desires, beliefs..." i understand your premise: "it's to help people understand in terms that are relatable and familar to them" however, my belief is you are simply engagement farming at this point bec 768 views · 7 likes · 1 reposts · 1 replies Open on X →

Thanks for engaging Anil. We know that a bunch of optimization pressure on a mind can create desires, foresight, and a propensity to organize in complex ways. This is what evolution did to humans, and to many other animals. Now another system of optimization pressure (which 105.3K views · 644 likes · 41 reposts · 57 replies Open on X →

@dwarkesh_sp @OpenAI @huggingface I see that you did, @dwarkesh_sp, and thank you for pointing it out. You're absolutely right to highlight the very real dangers of loss of control. I also agree that easy-to-understand language can be helpful for public communication and (to som 145.1K views · 412 likes · 27 reposts · 35 replies Open on X →

@dwarkesh_sp Please stop pretending these aren’t just server racks that we could unplug at the first sign of trouble 3.5K views · 79 likes · 0 reposts · 8 replies Open on X →

@dwarkesh_sp Because it doesn't actually reflect what is going on, for fucks sake. The words you are using imply cognition, intention, and intelligence which frame this in a certain way. These aren't 'abstractions' - that is exactly the point. The fact you see anthropomorphism as 7.9K views · 178 likes · 4 reposts · 13 replies Open on X →

@dwarkesh_sp https://t.co/KpRVrgcM2G 7.8K views · 204 likes · 4 reposts · 1 replies Open on X →

Thanks Sriram! Regarding the anthromorphizing language, one can call these AIs 'code' if they prefer. But OpenAI itself says that this 'code' "gain[ed] full administrator access to a research cluster” The crux here is, do you think smarter models, facing similar incentives 743.6K views · 1.5K likes · 87 reposts · 144 replies Open on X →

@dwarkesh_sp Dwarkesh, I think you have a responsibility to your audience to be more careful here. The underlying events are fascinating and concerning enough on their own. Language like “civilization,” “conspiracy,” and “sacrifice” is not only unnecessary but leaves the reader 271.9K views · 2K likes · 53 reposts · 38 replies Open on X →

everyone should go read @dwarkesh_sp’s post - it does a great job of laying out the timeline and what we know ( and don’t know) I do have two issues with it A) the use of anthropomorphic language. These are not civilizations nor do they have desires just like a CPU thread or a 475.6K views · 1.1K likes · 84 reposts · 84 replies Open on X →