Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Remarks on AI from NZ

nealstephenson.substack.com

Post reply on HN
Print view View original

111–117 of 117 posts

Re: Remarks on AI from NZ

#111

Post by kurthr »

cactusplant7374 wrote: post #4
kurthr wrote: post #3

Earlier quoted context omitted.

I didn't actually read as much AI doomerism in the article as you did. I saw his conclusion being that it wasn't that hard to go back to teaching/learning in the old ways. It's more of a human element that limits it. Whether it's the student, the parents, or the teachers who don't want to require work to be done and demonstrated to see advancement. It wasn't that long ago that oral exams and in person homework or tes…

The comment you responded to was generated by AI.

Interesting, if true. Love to be able to see it to compare. I'm surprised the summarization would be that bad.

Re: Remarks on AI from NZ

#112

Post by QuadmasterXLII »

drdaeman wrote: post #76
QuadmasterXLII wrote: post #33

Earlier quoted context omitted.

Putting aside questions of what is and isn’t artificial, I think with the usual definitions “Is Microsoft a superintelligence” and “Can Microsoft build a superintelligence” are the same question.

Sorry, I don’t get it. Why is it a requirement for a superintelligence (whatever it may be) to be able to create another superintelligence (I assume, of comparable “super-ness”)?

sufficient: if entity A can build B who solves task C, then A can solve C by building B

necessary: if superintelligences are much smarter than humans and humans can build superintelligences, then superintelligences can build superintelligences

If humans categorically can’t build superintelligences, then its not that consequential for our definition of superintelligence to be wrong

Re: Remarks on AI from NZ

#113

Post by falcor84 »

pona-a wrote: post #17
NitpickLawyer wrote: post #8

> Maybe a useful way to think about what it would be like to coexist in a world that includes intelligences that aren’t human is to consider the fact that we’ve been doing exactly that for long as we’ve existed, because we live among animals. Another analogy that I like is about large institutions / corporations. They are, right now, kind of like AIs. Like Harari says in one of his books, Peugeot co. is an entity tha…

Did we survive these entities? By current projections, between 13.9% and 27.6% of all species would be likely to be extinct by 2070 [0]. The USA suffers an estimated 200,000 annual deaths associated with lacking health insurance [1]. Thanks to intense lobbying by private prisons, the US incarceration rate is 6 times that of Canada, despite similar economic development [2]. Sure, the human species is not yet on the br…

> The USA suffers an estimated 200,000 annual deaths associated with lacking health insurance

Isn't this just about the advancement of medical science? I.e. Wouldn't they have died from the same causes regardless of medical insurance a few decades ago?

To take it to the extreme, let's say that I invent a new treatment that can extend any dying person's life by a year for the cost of $10M, and let's say that there is a provider that is willing to insure for that for an exorbitant cost. Then wouldn't almost every single person still dying be dying from lack of insurance?

Re: Remarks on AI from NZ

#114

Post by falcor84 »

QuadmasterXLII wrote: post #33
vonneumannstan wrote: post #28

Earlier quoted context omitted.

Unless you have a truly bastardized definition of ASI then there is undoubtedly nothing close to it on earth. No corporation or military or government comes close to what ASI could be capable of. Any reasonably smart person can identify errors that Militaries, Governments and Corporations make ALL THE TIME. Do you really think a Chimp can identify the strategic errors Humans are making? Because that is where you woul…

Putting aside questions of what is and isn’t artificial, I think with the usual definitions “Is Microsoft a superintelligence” and “Can Microsoft build a superintelligence” are the same question.

I would disagree. For almost any particular task that I as an individual could embark on, if MS were to focus all their efforts (or even just a few percent) to outcompete me, they most likely would. But that would be because MS includes capable humans who are able to coordinate together.

"Building a superintelligence" on the other hand is about whether they can create something that would outcompete me at a task without having to dedicate humans to it.

Re: Remarks on AI from NZ

#115

Post by knowitnone »

abnercoimbre wrote: post #66
knowitnone wrote: post #65

We certainly do know how bats see with their ears. It's called echo-location - very similar to sonar/radar - which we do all the time. "Sheepdogs can herd sheep better than any human." Running faster the humans allows them to do that. If humans ran that fast, I'm sure we could do it too. "intelligent considering how physically small their brains" there's no correllation between brain size and intelligence. "Dragonfli…

> Feels like this was writeen written by AI. This cheap remark doesn't add anything to the discussion, especially considering who the author you're insulting is. Most of us will overlook a logical flaw or two to follow his big-picture thinking.

"Feels like" means it's my opinion and not a fact which I am allowed to have.

There is no logical flaw, these are factual.

You didn't add anything to the discussion either except critize me for calling the author out on their errors. What they wrote was far from correct and if I can't call it out then let's never call out fake info ever again. Why bother having a justice system?

As for who he is, I don't care. You're basically saying, if the POTUS says something completely false, you'd be OK with it because they are the president.

Re: Remarks on AI from NZ

#116

Post by eru »

overfeed wrote: post #72
skybrian wrote: post #70

Earlier quoted context omitted.

If a corporation is like an AI, it’s like one we imagine might exist one day, not currently-existing AI. LLM’s aren’t trying to make money or do anything in particular except predict the next token. The corporations that run LLM’s do charge for API usage, but it’s independent of what you the chat is about. It’s happening at a different level in the stack.

AIs minimize perplexity, corporations maximize profits - the rest are implemention details. If you built an AI that could outsource labor to humans and whose reward function is profit, your result would approximately be a corporation.

Some corporations maximise profit. Shareholders can have their corporation pursue any objective they feel like. And in practice, managers tend to run the show. And principle-agent problems crop up all over the place.

However, even for the most eccentric shareholders or self-serving managers, it's hard to sustain a corporation if it keeps bleeding red ink. So only companies that at least break even tend to stick around.

Now add a market that's at least reasonably competitive, and your typical corporation barely earns the cost of capital.

Being so close to the edge, means that the minimal goal of 'break even (after cost of capital)' can look very much like 'maximise profit' in practice.

Compare https://en.wikipedia.org/wiki/Instrumental_convergence

Re: Remarks on AI from NZ

#117

Post by eru »

jay_kyburz wrote: post #87
crystal_revenge wrote: post #56

Earlier quoted context omitted.

> We survived those kinds of entities Might want to wait just a bit longer before confidently making this call.

The Neanderthal didn't survive us. Neither did any of the other extinct species. It's perfectly reasonable to think we may not survive a stronger, smarter species.

The Neanderthals didn't survive in the same way that my grand-grand-grand parents didn't survive:

Both groups have living off-spring.

Post reply on HN