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

After the AI boom: what might we be left with?

blog.robbowley.net

Post reply on HN
Print view View original

471–480 of 489 posts

Re: After the AI boom: what might we be left with?

#471

Post by esafak »

mrcwinn wrote: post #320
coderenegade wrote: post #164

Earlier quoted context omitted.

Nothing in capitalism suggests that consumers have to be human. In fact, the whole enshittification trend suggests that traditional consumers are less economically relevant than they've ever been.

Actually you made me think of a fun scenario. A robot, who I will name Robort, over time becomes, say, 1/10th the price of human labor. But they do 10x the work quality and work 10x longer. In that scenario, you could pay them the same wage but produce significantly more economic value. The robot, who won’t care about material possessions or luxuries, could make purchases on behalf of a human - and that human, overwo…

The company that built the robot simply takes the money that would have gone to the human workers. The robots will exert strong downward pressure on wages. Humans will have to look for jobs the robots can't do, and when they fail taxes will make up the difference, else people will revolt.

Re: After the AI boom: what might we be left with?

#472

Post by maplethorpe »

ACCount37 wrote: post #431
maplethorpe wrote: post #364

Earlier quoted context omitted.

> Hallucinations are incredibly fucking overrated as a problem. They are a consequence of the LLM in question not having a good enough internal model of its own knowledge, which is downstream from how they're trained. Plenty of things could be done to improve on that - and there is no fundamental limitation that would prevent LLMs from matching human hallucination rates - which are significantly above zero. Why is th…

To avoid hallucinations, you, a human, need two things: you need to have an internal model of your own knowledge, and you need to act on it - if your meta-knowledge says "you are out of your depth", you either answer "I don't know" or look for better sources before formulating an answer. This is not something that's impossible for an LLM to do. There is no fundamental issue there. It is, however, very easy for an LLM…

I read the page you linked, but I'm still not understanding why hallucination isn't an inevitability of LLMs. The explanation OpenAI gives doesn't feel like a complete answer to me.

OpenAI claims that hallucination isn't an inevitability because you can train a model to "abstain" rather than "guess" when giving an "answer". But what does that look like in practice?

My understanding is that an LLM's purpose is to predict the next token in a list of tokens. To prevent hallucination, does that mean it is assigning a certainty rating to the very next token it's predicting? How can a model know if its final answer will be correct if it doesn't know what the tokens that come after the current one are going to be?

Or is the idea to have the LLM generate its entire output, assign a certainty score to that, and then generate a new output saying "I don't know" if the certainty score isn't high enough?

Re: After the AI boom: what might we be left with?

#473

Post by ACCount37 »

maplethorpe wrote: post #472
ACCount37 wrote: post #431

Earlier quoted context omitted.

To avoid hallucinations, you, a human, need two things: you need to have an internal model of your own knowledge, and you need to act on it - if your meta-knowledge says "you are out of your depth", you either answer "I don't know" or look for better sources before formulating an answer. This is not something that's impossible for an LLM to do. There is no fundamental issue there. It is, however, very easy for an LLM…

I read the page you linked, but I'm still not understanding why hallucination isn't an inevitability of LLMs. The explanation OpenAI gives doesn't feel like a complete answer to me. OpenAI claims that hallucination isn't an inevitability because you can train a model to "abstain" rather than "guess" when giving an "answer". But what does that look like in practice? My understanding is that an LLM's purpose is to pred…

The answer is that it does know. Not exactly, but the "general shape" of the answer is known to the LLM before the very first token of the answer is emitted!

"Next token prediction" is often overstated - "pick the next token" is the exposed tip of a very large computational process.

And LLMs are very sharp at squeezing the context for every single bit of information available in it. Much less so at using it in the ways you want them to.

There's enough information at "no token emitted yet" for an LLM to start steering the output towards "here's the answer" or "I don't know the answer" or "I need to look up more information to give the answer" immediately. And if it fails to steer it right away? An LLM optimized for hallucination avoidance could still go "fuck consistency drive" and take a sharp pivot towards "no, I'm wrong" mid-sentence if it had to. For example, if you took control and forced a wrong answer by tampering with the tokens directly, then handed the control back to the LLM.

Re: After the AI boom: what might we be left with?

#474

Post by senectus1 »

CuriouslyC wrote: post #84

This is where we're headed: https://sibylline.dev/articles/2025-10-12-ai-is-too-big-to-f...

what a terrible article..

Compares China's current 2025 energy capability with US at the end of 2023. Then starts throwing 2025 uptake numbers for solar and wind into the 2023 numbers as if having a TWO YEAR GAP in the numbers dont matter at all.

Re: After the AI boom: what might we be left with?

#475

Post by nebula8804 »

kentm wrote: post #458
nebula8804 wrote: post #397

Earlier quoted context omitted.

I don't really know. I think a lot of them migrated off of Bush. I don't have the clip right now but I had old interview videos showing how high and mighty Bush supporters were and how they supported the Iraq war all in going into his second term. They went so far as to act like Trump supporters against anyone who didn't support the "commander in chief" (ie. those people should be put in jail or kicked out). The GFC…

As evidence of this point, “Trump Derangement Syndrome” was originally “Bush Derangement Syndrome”. I have to roll my eyes anytime someone mentions TDS seriously. Although Trump has a cult of personality that Bush did not have; you didn’t have people adorning their lawn with Bush gear in the same way as Trump.

It was a different era. Bush and the old Republican guard were never imaginative nor were they product people. Trump started with his signature red hat but look at what it blossomed into.

No one should have been surprised. Trump has a long history of being a product person (Trump Steaks, Trump University, Trump Vodka, Trump Fragrance, Trump Mattresses, Trump: The Game etc.)

You also have to consider that the pre Web 2.0 world of the internet during the Bush years also had less opportunity of flooding the market with branded crap. You had to call a phone number or mail out a money order to order merch.

Re: After the AI boom: what might we be left with?

#476

Post by nebula8804 »

CuriouslyC wrote: post #418
nebula8804 wrote: post #398

Earlier quoted context omitted.

The south is pretty modern all things considered. You wanna see real pain? Go ask Chinese people who had to sacrifice these past few decades to make China the powerhouse it is today. Its the reason why they can tolerate anything the US throws at them. Their pain threshold is much higher than whatever the west has. These southerners could have lost their culture, the local industry has long since left, and fentanyl is…

I'm a southerner, and I road trip a lot. Driving through parts of the southeast you will pass through a sea of dilapidated trailers with yards full of junk, run down chain restaurants and industrial lots. It's not India or the Philippines, but it's still pretty third world in a lot of ways.

Of course its third world in some ways, but take those Chinese people I spoke about, swap them into the same spot as the people currently living there and I GUARANTEE you that they will turn those places around and probably make a whole super competitive region that could go toe to toe with any of the urban areas of the country.

It goes back to my point. These people in those areas think they are in for bad times but they have not seen REAL pain yet and yet they are playing games right now because they think they have nothing to lose.

Imagine if your theory comes true and the government has to cut services to the bone. No more medicare/medicaid/maintaining their infrastructure(however shit it is now it gets worse)/state subsidies for states like Nebraska.

There are probably horrors that im not even thinking of.

Re: After the AI boom: what might we be left with?

#477

Post by jhanschoo »

jononor wrote: post #408
jhanschoo wrote: post #391

Earlier quoted context omitted.

A pillar of the author's argument is that > Chinese will not pull back on AI investment, and because of their advantages in robotics and energy infrastructure, they don't have to. > The gap matters because AI's value capture depends on having the energy and hardware to deploy frontier models in the physical world. I understand energy infrastructure because of power-hungry GPUs and ASICs, but I was wondering about the…

Early day for robotics and AI still. But for a superpower, the potential for (more) autonomous weapons in that intersection is something that (unfortunately) cannot be ignored.

I have the knowledge of a layman in robotics, but rather that autonomous control systems, I'm more curious about the prospect of improving world models in AI by sensor fusion at an unprecedented scale, addressing the limitations of today's models only having empirical knowledge pretty much second-hand from literature.

Re: After the AI boom: what might we be left with?

#478

Post by maplethorpe »

ACCount37 wrote: post #473
maplethorpe wrote: post #472

Earlier quoted context omitted.

I read the page you linked, but I'm still not understanding why hallucination isn't an inevitability of LLMs. The explanation OpenAI gives doesn't feel like a complete answer to me. OpenAI claims that hallucination isn't an inevitability because you can train a model to "abstain" rather than "guess" when giving an "answer". But what does that look like in practice? My understanding is that an LLM's purpose is to pred…

The answer is that it does know. Not exactly, but the "general shape" of the answer is known to the LLM before the very first token of the answer is emitted! "Next token prediction" is often overstated - "pick the next token" is the exposed tip of a very large computational process. And LLMs are very sharp at squeezing the context for every single bit of information available in it. Much less so at using it in the wa…

By "shape" of the answer, what do you mean? I always visualized token prediction as a vector pointing off into some sort of cloud of related tokens, and if that's a fair way to visualize it, I could understand how you could say, before even emitting the first token of the answer, "we are pointing towards generally the correct place where the answer is found". But when a single token can make or break an answer, I still don't see how you can truly know whether the answer is correct until the very last token is reached. Because of this, I'm still not convinced hallucination can be stopped.

Can you help correct where I'm going wrong?

Re: After the AI boom: what might we be left with?

#479

Post by bpt3 »

discordance wrote: post #330
bpt3 wrote: post #222

Earlier quoted context omitted.

> brutal serfdom You can't seriously believe that spending all your income each month while living in the country with the highest standard of living in history is "serfdom." Hyperbolic nonsense like this makes the rest of the article hard to take seriously, not that I agree with most of it anyway.

The US is ranked 14th in terms of standard of living in (in order): Luxembourg, Netherlands, Denmark, Oman, Switzerland, Finland, Norway, Iceland, Austria, Germany, Australia, New Zealand, Sweden, United States, Estonia. Based on these metrics: Quality of Life Index, Purchasing Power Index, Safety Index, Health Care Index, Cost of Living Index, Property Price to Income Ratio, Traffic Commute Time Index, Pollution Ind…

And by a very objective metric, which is disposable income for the median household, the US is #1.

Sorry if I don't take numbeo.com all that seriously. If you do, you should poll people to see if they'd rather live in Oman or the US and think again.

Re: After the AI boom: what might we be left with?

#480

Post by dcminter »

notepad0x90 wrote: post #355
dcminter wrote: post #174

Earlier quoted context omitted.

The dot com crash was a thing though. The bubble burst. It's just that there was real value there so some of the companies survived and prospered. Figuring out which was which was absolutely not possible at the time. Not many people foresaw Sun Microsystems as being a victim and nor was it obvious that Amazon would be a victor. I wouldn't bet my life savings on OpenAI.

Google, Microsoft, Apple, the value it's generating for them alone is enough of a demand to refute the "bubble" claims. the hype bubble will surely burst, but people are using it everyday. This is like saying the "crack cocaine" bubble will burst once people hear about its side effects (or alcohol lol). A reduction, sure. But not a crash. I wouldn't be my life savings on OpenAI either FWIW.

People were using Amazon every day. But people were also using Pets.com every day. A bubble bursting doesn't make everything disappear. Just some things.
Post reply on HN