Hacker news

  • Top
  • New
  • Past
  • Ask
  • Show
  • Jobs

My Empire of Dirt (https://geohot.github.io)

136 points by throwaway384034 3 days ago | 179 comments | View on ycombinator

ablation 3 days ago |

Slightly worried about geohot given this and other doom-laden posts. He feels a little terminally online.

flyingsky 3 days ago |

>But it leads to World War III.

No! I'd expect a tense standoff around compute and Taiwan. Not a world war.

If a first strike can’t outthink drone production, the domination attempt just fails. If China wins any war longer than a year and is pulling ahead on electricity, it has no reason to escalate.

The closest precedent is the 1945–49 US nuclear monopoly. Preventive war was seriously argued for and never launched,the other sides raced instead.

hypfer 3 days ago |

This reads not too dissimilar to the last post of his that I saw end up here and being flagged soon after.

I don't buy the world war theory. I do expect unrest regardless of AI outcomes, but that will be more internal than between countries.

But at this point that's just reiterating what has been reiterated many times with zero new insights.

saberience 3 days ago |

What?

So, with no reasoned argument, this guy somehow predicts the end of the world is coming and its effectively inevitable?

And people on this site take this couple of paragraphs at face value because... no real reason?

The Second World War didn't end the world, neither did the Cold War, and the world is now far more economically and politically intertwined than it ever was.

Do you think Russia really wants to nuke London? Do you know how many Russian oligarchs have apartments and houses in London where their extended families live? Similarly, do you know how many Chinese people live in London or how much Chinas economy is tied to Europe and the US?

If the world didn't end in the Cold War, it certainly isn't going to end now, and especially over AI, which although a cool tech, is incredibly far from this mythical singularity. Have you seen modern robots? Have you seen LLMs trying to do anything that isn't easily verifiable (with software)? E.g. try and create new drugs? Or design new materials, machinery, or complex systems like a fusion reactor.

Most people on this site are programmers and as such they are in a tiny bubble, obsessed with AI now its equal to them at the very act they have depended on for their entire careers.

But no, the end of the world is not coming any time soon. The rich Americans want to stay rich, the rich Russians want to stay rich, and the rich Chinese want to stay rich, and bombing each other and dying in fire isn't part of their agenda.

ford 3 days ago |

The assumption here is that the US would go to war to keep China from becoming the global superpower

I personally don't want to live in a world where China is the global superpower. I also think nuclear war is a worse alternative (and I think the US agrees) - but I hope Western countries understand they need to build (energy, civil infrastructure, compute, manufacturing, etc.) if they don't want to pick between these two options.

undefined 3 days ago |

undefined

grim_io 3 days ago |

Maybe the negativity comes from seeing local AI falling further behind rather than catching up.

g-b-r 3 days ago |

I didn't know how close Amodei is to EA (although stating to not be part of it)

simianwords 3 days ago |

This guy is such a grifter.

I was downvoted 5 months back when I commented on a post he wrote called Eternal Sloptember where he described that coding agents would never get better. I made a bet that this guy would change his mind. Not only did I get downvoted, I had like 5 people trying to convince me that I'm wrong. Epistemic hygiene should exist.

This is what he wrote btw

    I’m calling it now, the adoption of AI agents into software development will be one of the most costly mistakes in the field’s history. Agents cannot program, and it’s taking longer and longer to realize that they can’t. 
Here’s my comment:

https://news.ycombinator.com/item?id=48263483

Now he’s suddenly doomer posting. Just 5 months back he said AI couldn’t code.

I would be very cautious of a person who writes this way

undefined 3 days ago |

undefined

markus_zhang 3 days ago |

I kinda agree that WW3 is actually better than perpetual dark age of cyberpunk, if it gets to that bad.

fwlr 3 days ago |

  The future doesn’t need suppliers. The future doesn’t need customers... The future definitely doesn’t need you. After RSI comes, all the demand will be internally generated, agents working with other agents, everyone else cut out of the loop
Yeah, this is the part I wish more people would consider. I see a lot of debate about “intrinsically human production” vs “building AI to be a better producer” and the like. This is stuff like “it’s not art unless a human drew/wrote it”, anything to do with the term ‘slop’, all the way through to “AI is a better programmer than me” and “all junior developer roles have been taken by Claude”. On the means of production in general, the trend seems to be a God-of-the-rapidly-shrinking-Gaps rearguard fighting retreat as AI steadily gains ground.

I see essentially no discussion of the other half of the equation, what we might call “the means of consumption”. Everyone is operating with a default unexamined assumption that consuming is intrinsically human, that “consumers” and “people” are synonyms. The closest anyone seems to get is the “who will buy your stuff if we don’t have jobs anymore to pay for it?” -> “universal basic income” -> “from all those famously generous tax-paying capitalists?” cul-de-sac.

Nobody’s talking about, what if AI can just build a better consumer, too? When humans are left without an ecological niche in the economy, what then? Like the economic equivalent of a flightless bird on a remote island with no natural predators encountering an invasive species for the first time, we would go the way of the dodo.

busssard 3 days ago |

the author might enjoy the movie Zero Theorem...

qwertyforce 3 days ago |

if there is a possibility of decapitating first strike, then we need to have a Perimeter system

jaco6 2 days ago |

[dead]

josefritzishere 3 days ago |

[dead]

micahdittmar 3 days ago |

The blog hits on a real tension: literal-minded benevolence can easily slip into safetyism, domesticating us like in Prime Intellect.

That said, I think there’s a spectrum we're missing when we frame the future purely out of fear. Instead of letting AI strip away all friction, we can design guardrails that use it as a catalyst for personal growth, freeing us up to choose the things we actually want to struggle against.

marsven_422 3 days ago |

[dead]

micromacrofoot 3 days ago |

It's not hard to see from his perspective, if frontier AI research is right and not just a front for marketing, by their own words we're building the most unstable doomsday weapon imaginable. Why would they be able to control it?

the best outcome in this case is general human incompetence

moolcool 3 days ago |

I'm not scared that the AI companies are summoning some earth-shattering angel of death.

What does worry me a bit is that the people running these AI companies _think_ they're summoning some earth-shattering angel of death, but they are so self-righteous (literally) that they think it's actually good that they're doing this.

science4sail 3 days ago |

This is definitely geohot anti-Anthropic (misanthropic?) doomerism, but it's nice to see that people are finally realizing the answer to the question: "what will happen to the economy if no one has jobs".

The answer: there still will be an economy, you just won't be a meaningful part of it.

BaseBaal 3 days ago |

Was hoping this would somehow relate to Nine Inch Nails... Now I'm hurt.

Papazsazsa 3 days ago |

Doomers to the left of me, scammers to the right

Here I am, stuck in the middle with you

bob1029 3 days ago |

> Have you ever played an RTS game?

I have observed that the humble photon cannon rush is very effective. If not spotted it's usually a really bad time. Long term, is pretty obvious what happens here. If we get there.

Who dominates the gravity well that we all exist inside right now?

boriskourt 3 days ago |

RSI Repetitive Strain Injury? When it comes, nobody survives.

N_Lens 3 days ago |

Nice bit of doomerism from george.

tharien 3 days ago |

'The Metamorphosis of Prime Intellect' doesn't take into consideration any of the intuitive drives of humanity. Once basic needs are met, we don't become cat's waiting on the next meal. The utopia described in the book appears to be from the perspective of either someone very boring, or only able to get joy from the misery and strife of others.

smashandgrab 3 days ago |

I don’t think I can name one tech celebrity that I know of that hasn’t gone off the deep end either way when it comes to AI.

Even names I recognize on HN seem to either think AI is the 5th horseman or some benevolent intelligence coming to solve all of the worlds problems.

I think these people need a dose of ennui.

throwaway----- 2 days ago |

My name is george hotz but everbody calls me geohot <insert Daft punk - Giorgio>

dangerface 3 days ago |

Don't drink the cool aid guys.

navane 3 days ago |

The scramble for AI has many parallels to the scramble fro africa.

When one side is on the verge of winning everything, perceived or real, the other side will just pull out higher risk/reward strategies. Yes I also play RTS.

I can imagine many doom scenarios but I just don't see a single positive. The AI's will do our jobs and our overlords will give us free food and housing? Keeping us around as pets? Out of sheer good will and because of the cuteness of our eyes?

LogicFailsMe 3 days ago |

Dude needs to touch grass. The future is going to be a bumpy ride, but so is the present, and so was the past.

That said, I think the EA people need to touch grass too.

And while I don't think much of people myself, I don't think that immediately leads to WW3.

childintime 3 days ago |

He clearly sees capital at work up close, so it's likely no surprise he sees doom better than the rest of us.

vkgcr 3 days ago |

If RSI does takeoff and bring about the AIpocalypse, then at least the EA cult will have a decent chance of becoming a real religion!

senordevnyc 3 days ago |

I grew up reading the same LessWrong posts

Ah, so that explains the super weird culty vibe of this blog post.

duggan 3 days ago |

Think this chap needs to go touch grass.

aanet 3 days ago |

I, too, enjoy a bit of doomer sci-fi now and then...

/s

On a more serious note, it's interesting to see people's perspectives change and evolve, but not their opinion about their own right-ness -- which, as the tautology goes -- is always right.

FLeXMurphy 3 days ago |

I don't see the doomerism in his tone that you're all bringing up. He's taking shots at EA and accusing Anthropic and others of bullshit doomer marketing that could have dire consequences if they don't stop LARPing.