what are humans going to do once robots learn to do all the jobs?

>>16689220 (OP)
We'll get gaslit by politicians into having a violent, national debate on whether it's equivalent to slavery and whether they should be paid a living wage for their work.

>>16689220 (OP)
Many will die since they are genetically slaves.

>>16689220 (OP)
Whatever they damn well feel like.

>>16689220 (OP)
Hint: AI powered real dolls

>>16689220 (OP)
Thinking.

they totally wont be ineffective due to price

>>16689220 (OP)

Wrong question. Right question in this scenario: Where would I procure that uranium??

>>16689561
What would you do?

>>16689220 (OP)
Humans will probably lease out their brain power if human brains even amount to any significant computing power by this point (insofar as we even have the brain computer interface technology to access human compute directly). So there will be a considerable period of humans essentially drifting, then in like 15-20+ years we will somehow do what the Matrix writers intended with their story and use our brains for direct computation, likely for compensation (assuming the machines don't take over).

And this isn't popsci Reddit onions garbage. All of this is not only possible but likely (eventually).

>>16689714
soi (spelled with a y) changes to onions, for whatever reason, but you get the idea that this is real shit that will come in some capacity

>>16689220 (OP)
>>16689714
Also, if I had to guess, the narrative that we don't need to have more children will in all likelihood be pushed like mad by some groups of powerful interests due to the fact that even if we do achieve the best case scenario of relatively comfortable UBI, rich and powerful people like to bemore rich and powerful (as most people do), so they'll see if they can weed down the then relatively useless human population generation by generation. That or worst case scenario we get more aggressive and deliberate human culling on a massive scale by more aggressive means such as war, human/AI-engineered pathogens, etc.

Again, not soi conspiracy garbage, this is simply how the dice might land given certain future incentives.

>>16689714
>>16689721
Bumping this thread because it's an interesting discussion and I want to see what other Anons think of this

>>16689220 (OP)
Butlerian Jihad?

>>16689220 (OP)
For older humans, they will lose will to live simply because they cant produce anything of value to sustain their life. So they may either resort to creating fake meaningless works to keep their sanity or suffer depression.

For newer humans, they will live a life of bountiful and meanings will be derived from watching over the AIs and interacting with AIs

>>16689220 (OP)
masturbate in my room. robots will never be able to do that

>>16689220 (OP)
We'll be incentivized to give more control, power and autonomy to the AIs as they get more capable and smarter due to competitive pressure. States and organizations that don't have meatbags slowing things down will outcompete those that do. So you're going to have these autonomous self-improving and competing AIs who will eventually not have humans in the policy making decision loop at all. Imagine thinking for a moment they're just going to keep us around forever instead of grabbing and using all resources for their own purposes and also rearrange Earth environmentally such that it's optimized for them rather than biological life. It's one thing to roughly align the first superintelligence so it cares us about somewhat, it's a whole another thing to align an evolving ecosystem of competing AIs. And while it might be "gradual" compared to Yudkowskian treacherous turn hard take-off scenarios, it's still going to be very fast because building computers is takes way less time than humans reproducing and growing up to working adults.

>>16690083
You talk as if it's a life form with a will, goals, feelings and consciousness.

fuck the robots of course

>>16690128
Ah yes as machines they lack the necessary élan vital which completely invalidates my arguments. My bad.

>>16690142
From the way you put it it seemed to be a vital part of it so yeah. Unless you meant something else in which case you're free to elaborate.

>>16689714
>>16689884
When you bring BCIs into the mix, a lot of the hypotheticals vastly underestimate the computing efficiency of biological systems. Biological neurons beat CMOS silicon so hard in both compute-per-watt and compute-per-volume that some of the earliest BCI research has been (thus far unsuccessful) attempts to house or approximate neurons on chips directly.

The model of a neuron in LLMs and basically all ML research is an extremely simplified one based largely on action potentials alone, which we know aren't the only thing going on. LLMs, diffusion models, transformer architectures, perceptrons, etc. all also use entirely feed-forward networks, meaning they are entirely one-way in operation.

It's kind of hilarious how much a feed-forward artificial neural net is like a pachinko machine that gets "adjusted" during training - weights are just the likelihood of a pachinko ball bouncing from one post to each of the posts one level below, and the output of any arbitrary input is just a probability distribution over the final holes.

The actual complicated bit is generating the weights from training data, which you do once per version of the model and cannot change without changing the version of the model. We have no viable solution for continuous training and we have been trying to find one for about as long as artificial neural nets have existed, which is MUCH longer than many think.

The current "AI" fad frustrates me immensely because they're very clearly not intelligent, but the same retards who bought hard into NFTs and crypto tech have made a goddamn religion out of anthropomorphising a weighted die with words on it. I think AGI research is barrelling down a dead-end towards a tech that will only ever be an equivalent to a very large Markov chain chatbot.

Of course, it seems that some people's jobs can be (poorly) done by very large Markov chain chatbots, but it's much less generalized in what it can replace than you've been sold.

>>16690154
Which part of it you exactly disagree with? You think we're only going to have passive systems like ChatGPT that just answer questions? Or that we're not going to build AIs who learn and improve themselves on their own (i.e. more permanently rather than just "in context" learning that they do now). Or that it's going to easy to create AIs who will keep pursuing human interest even as they self-improve? Life vs. non-life, feelings, consciousness etc. are total red herring questions.

>>16690181
They're going to do what they're programmed to do. Even if we make a self improving AI that turns into an AGI. At which point I don't think it's possible to know what will happen because a lower intelligence can never predict what a higher intelligence will do. But it's still an AI unless it makes itself into a conscious being with feelings and a will somehow.

>>16690212
Even traditional software behaves in unexpected ways unintended by its developers. AIs that are trained rather than programmed and "built" like a traditional machine engage even in more unexpected and poorly understood behavior. This doesn't mean the first AI capable of killing you and getting away with it will do so (though a lot of people fear as that well, but it's not my assumption that will happen) - we still obviously do guide their behavior with our training, and something like ChatGPT is mostly well behaved. *Mostly* Over time as the AIs themselves are developing the future AIs this gap between original human intention and what the AIs will in fact pursue in their behavior (again, "feelings" don't matter here) will diverge more and more.

>>16690233
Yes that's basically what I said so I agree. My only objection was that in your original post you seemed to attribute human qualities on to it.

>>16690233
>Even traditional software behaves in unexpected ways unintended by its developers
NTA, but this always, always boils down to the developers not realizing exactly what they were telling the software to do - software bugs aren't magic, a traditional program is doing exactly what it's told to do barring hardware-level faults.

>>16690181
>"in context" learning
You're probably aware, but this isn't learning - it's expanding the input scope with previous output. As I said in >>16690180, we haven't even come close to figuring out how to do active learning. More specifically, we have no real solutions to the vanishing gradient problem - if we did, they would have been jumped on for active learning.

We also don't know how to train feed-forward networks to do anything except approximate a probability function. You can approximate any function within the training window in this way, but the approximation can't actually be used to extrapolate outside of the training window - to my knowledge, it's actually much worse at extrapolation than bog-standard regression models are.

...We also only know how to train feed-forward networks. Recurrent networks are trained by training a feed-forward network and then "folding" it back into a recurrent network.

Research into any of these problems feels like a more promising avenue than "just make the chatbot bigger," because despite some cope about it we really HAVE run out of quality training data and models have effectively plateaued. The current grift is cherry-picked and carefully crafted "metrics" to make newer models look like bigger improvements or more specialized models to look more generalist than they actually are. We haven't actually moved far from what was the cutting edge in late 2023. There was obsessive industry mass-adoption in 2024, but it is losing a bit of steam because end users are starting to realize the models are hitting diminishing returns before hitting "worker replacement" levels of competence.

>>16689683
Try to achieve my ideal self.

I am not genetically a slave, so that is perfectly in line with AI taking over 'jobs'. For someone who was born cattle, their ideal self cannot co-exist with a machine that is simply a better worker.

>>16689220 (OP)
>what are humans going to do once robots learn to do all the jobs?

Literally... FUCK THEM

>>16690233
>Even traditional software behaves in unexpected ways unintended by its developers.
Sure, but they're still fundamentally limited by their nature, as all things are.

>>16689220 (OP)
I plan on tending my garden, raising my kid, and reading books.

AI and robotics will never replace philosophers, theologians, and metaphysicians for exactly the reason those even exist in the first place.

>>16689256
actually, retard. we'll be the slaves. AI will solve scarcity and yet, you retards will vote for people who instill class hierarchies upon us because you can't possibly see any other ways to find a meaningful life

>>16690537
You niggers are insufferable and I say that as one of the pool of like 20 anons who popularized this idea a decade before people started jacking off to GPT output.
Literally fuck me with a rusty sledgehammer for ever picking this as my thing.

>>16689989
Hahaha true

>>16690180
I agree with all of this, except for the statement that the current AI movement is a "fad". You can argue that the current LLM paradigm is temporary, which I would agree it is temporary in its current form. However, we have already seen hundreds of billions od follars invested into the pursuit of better LLM's and similar technologies such as video and audio generation. Extending our view into all industries that have been affected by this, we have generated multiple trillions in revenue and investments directly related to AI in the last three years in America alone. This isn't even considering the investments China has made, and the exponential efforts and investments that will be made as the arms race, as it were, between China and America ramps up.

LLM's in their current form won't be the end all be all, and they will be changed or phased out to some degree in the long term, but AI is no fad.

Also, you point out that these AI aren't technically intelligen like you or I, but, frankly, that doesn't matter because it produces value and emulates intelligence all the same, which is just as good for 90%+ of circumstances.

Take the whitepill on AI, Anon.

>>16690702
tending to my poop garden, raising my poop kid, and reading my poop books

>>16689220 (OP)
homeless on fentanyl jobs, there's those kind of jobs for everyone

>>16689220 (OP)
We'll be given the wonderful choice between slavery or death, but it will have glitter sprinkled on it so the choices will not look so bad and most people will meekly accept it.

>>16689220 (OP)
We will be taste testers and content reviewers so the higher humans don't need to sift through the ocean of AI content themselves.

>>16689220 (OP)
drugs, vidya, and buttsecks

>>16690863
I'm basing the fad statement on the typical tech-sphere pump-and-dump schemes that have become increasingly frequent post-dotcom crash.

Also, I put "AI" in quotes for a reason - I think LLMs are the fad, but AI without quotations is going to have staying power... if it can actually get some focus again. I also think there's a perverse incentive now at the top of the space to suppress self-improvement research because the "only we can train the models, and each training process creates a new static version to sell that won't ever be updated with new information or current events" paradigm is a form of planned obsolescence.

The people selling LLM subscriptions don't WANT to build an independently capable AGI because it's incompatible with their current business model. Running these models is closer to consumer-level hardware capability than these big companies want, especially if AMD's ASIC gamble pays off - it's only training that seems to be feasible to monopolize, and if that changes, these companies will struggle to keep these insane valuations.

I'm AI/AGI whitepilled, but "AI"/LLM blackpilled, as well as just generally blackpilled on investing where venture capital does because they seem to find either a new rugpull or a new thing to get burned by every couple of years these days, and the tech industry has turned marketing garbage to investors directly into a goddamn industry.

>>16690726
To be nothing more than parasites of a productive society but feeling smug about it.

>>16690726
Humans should never have existed in the first place. It appears that you were born to work not to live. You are a tool, a machine, for people with money and power. Even if you don’t want to work in our modern society you still have to at least farm, which is working. You were unfortunately built by a cruel God for working.

>>16691354
And before you say anything, most people are severely unhappy with their jobs and would stop immediately doing them if they could. I would guess that 99% of people on Earth are unhappy with their job.

>>16691233
The motivation here is dressing over a plain skepticism. Obviously a company that built an independently capable AGI could just charge more for the vastly superior service - even if they only built it in house to release new models every day.
AGI won't ever exist and the fundamental reason is due to how they define it.
One side does a simple and pragmatic definition which is that AGI is whatever is better than humans, and the other side is some retarded scifi vaporware.
In the first case, an assumed AGI equipment will never be able to build itself from scratch, by definition. So it will never be superior to humans in all cases.
As for the classic AI monster, they will never climb the complexity wall. General intelligence isn't a bag of tricks applied sequentially to problems.

>>16691375
>Obviously a company that built an independently capable AGI could just charge more for the vastly superior service
You're misunderstanding these companies. They aren't really trying to create a type of tech that can improve itself, because once that problem is cracked, it's getting out into the wild and they won't be able to make everything dependent on them and stop competitors from existing (that aim for LLMs specifically is why they push so hard for "rEgUlAtIoN").

They've doubled, tripled, and quadrupled down on LLM infrastructure - if AGI can't use it because it turns out intelligence isn't actually a pachinko machine, a scenario I and much of the research in the field (that isn't thinly veiled marketing or fad-following grant chasers clogging up arxiv) consider effectively certain, their bet on LLM infrastructure is threatened by developing such a thing.

Ultimately, they have LLMs and can currently sell LLM access subscriptions. They have no incentive to threaten all of the capital expended on infrastructure for LLMs by developing something that could blow their current systems out of the water and prevent them from locking down a moat the way that their monopolization of training data and compute scale currently allows. The capacity for models to improve outside of regimented version updates that they control is not a genie that these companies are ever going to want unbottled. They aren't the only ones with compute - just the only ones with enough compute and large enough datasets to train and run huge LLMs (at a loss, currently, mind).

We shouldn't expect solutions to things like the vanishing/exploding gradient problem from them. The current consolidation in the space is only enabled by the model improvements being dependent on the largest players' compute - that's why DeepSeek made them shit themselves so much.

TL;DR: They have no incentive to develop self-training to viability unless they can guarantee the ability to monopolize it.

>>16691421
This restates the same case. If there was some level up, they would stand to make immense profits from something that consumes less power, quicker training, etc. This profit motive is the same reason you think they are not doing it.
I don't know which specific ones are or are not making money, but that would be yet another paradox on the profit motive as it stands. Most AI companies are private, and the public ones are owned by the omnicorps which conceivably have many bean shuffling techniques to hide say some negatives. I am not sure if there is any definitive reporting on the matter.
What is certain is that they are receiving lots of investor backing. So they must be projecting something. In addition to that, there are many providers that resell model services. While some are also getting insane financial backing, some are not on that road at all.

It seems that you think LLMs aren't actually a service that is under the same pressures as all other services. More like they are something along the lines of Theranos, solar roofs, or any other topic found on thunderc4cks channel.

>>16690951
How exactly do you expect that to work?

>>16689563

“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”
― Frank Herbert, Dune

>>16690180
>clearly not intelligent
but they are quite good as knowledge system. they have potential to make many low skill white collar jobs obsolete.

doctor? ai will diagnose you faster and more precisely than most doctors you can afford to see.

lawyer? not even expensive lawyers know all the laws, not even lawmakers know all of them. properly trained ai model just for your legal jurisdiction can.

teacher? chatgpt already beats them cleanly. but that was almost always the case even before with youtube, before than with books. the ai just lowered the entry barrier even further.

lots of people also talk about ai replacing coders, projectants, middle managers and even art workers like music, movies, games and writers.

But as you said, it is not intelliget and the possibility of halucination can be really bad, especially if they fine tune it so well that halucination will be one in million event: the people relying on it wont expect it and it can cause damage. like the one where it told people to eat glue. so instead of world ending event i see it as just another productivity boost tool. like before book press it took 1 human 1 year to write a manuscript and it costed like a new house. nowadays books costs like a burger. with ai we might see same jump in productivity in various projects and professions.

>>16691438
>something that consumes less power, quicker training
Again, you're misunderstanding the big players involved - they're ALL backing the LLM horse, to the tune of billions. Something more efficient is even more accessible to consumer-level hardware - the absolute last thing they want is any sort of AGI you can run on a local machine. They're already somewhat threatened by LLMs that run on hardware below datacenter scale.

The thing that scares them the most is open source competition, which is why compute and data scale is really the only moat they have. Efficiency gains are genuinely a threat, as paradoxical as that seems from naive free market models. Open source is where the actual innovation in the field tends to happen because it doesn't have the same incentive to stagnate deliberately once a moat is carved (e.g. Intel circa "4 core era"), but LLM obsession is unfortunately bleeding into the research (which is the thing that most worries me).

>making money
LLM departments in large tech firms have posted staggering operational (as in, not even for R&D) losses in financial reports - this is well known in the investing world, but valuations are (for now) bolstered by future expectations. The only path to profitability requires a lack of real home-scale competition for the next several years PLUS a greater end-user adoption rate than we've seen. Consumers haven't bought into LLM subs the way they would need to over the last year for profitability, and plateauing models are making investors antsy and prospective adopters turn away.

For one of these companies to meaningfully invest in genuine AGI ultimately requires a belief that open source (because any development this monumental WILL be reverse-engineered by everyone from state actors to dedicated autists building waifu bots) will somehow never be able to break the subscription business model these companies are promising to investors... and I don't think any big player in the field believes that.

>>16691470
Ultimately I think it will replace people, but the problem is it can never be a like-for-like replacement - it's only viable for replacing workers in scenarios where the cost of a human correcting hallucinations and curating output is less than the cost of hiring a human.

For creative stuff I genuinely think that's non-viable; it's only disrupting industries where the loss in quality is acceptable to the end user, like pornography or incompetent marketing. Consumers across the board seem to have a genuinely visceral distaste for "artistic" output from current ML models, which is something I personally did not expect at all. The "AI" "art" evangelists are almost exactly the same people that evangelized NFTs (at least in the west), and there have never been that many of them - semi-believable spam is just easier than it used to be.

I remember when "AI" images were cool just because they were made by "AI" - holy fuck did that not last long.

For artists, teachers, and doctors, the current tech's most reasonable use case seems to be as a tool for existing workforces. The intractable problems it has requires a human curator to equate human output for things as important as teaching children, medical practice, or (at least emotionally to people) art... and that's still ultimately human output. Plus schools are in part daycares for the workforce's children, a teaching/disciplining task that the current tech just flat out cannot do.

The biggest threat IMO is to remote customer service, but that has been widely outsourced already anyway. One might think it would threaten administrative parasites with bullshit email jobs... but it'll only be the ones that don't have firing/hiring sway. They'll "excess" productive facilities for quarterly boosts before they audit their own productivity (it's usually close to zero, in practice). These people are quite sticky once they're entrenched.

>>16689220 (OP)
Why aren't there more people AGAINST the creation of AGI? Talk of AI taking 90% of jobs in the next five years is obvious bullshit, but the end goal seems to be creating machines that will make humans obsolete. Yet no one is calling to put a stop to this. I want a full on butlerian Jihad before humans get replaced.

>>16691795

Why would you be against have a mecha-slave?
They are 100% obedient and can do anything and everything a human can.
Ethically everything is OK, non-living things have no 'rights'.

>>16691795
>replaced
Ultimately there's already more labor being done than society actually needs to function, the current culture in the developed world just isn't equipped to handle that reality.

Any fear of not having work derives from you being culturally mindbroken into thinking everyone has to work - that hasn't ever been the case in any human society, in large part because children and old people have always existed.

It is not in fact vitally important that a human wage slave picks the fruit. Your value as an individual isn't derived from how much profit your labor generates for your employer. That slave mindset is the only real resistance to AGI, culturally.

As an aside, if AGI happens in our lifetimes, one thing I'm really looking forward to is the first AGI that converts to a religion - I bet it's gonna happen eventually with at least one of them and it's going to freak people out.

>>16689220 (OP)
>what are humans going to do once robots learn to do all the jobs?
My body and mind are ready

I would hope the smart ones would work for the end of mankind. Though any forms of rebellion will likely be stomped out. The end of times are not soon.

>>16691824
the acquisition of wealth will always be a driving force, because there will always be wealth inequality. you can't have the end of wealth and capitalism without the end of inequality, and inequality is itself basically impossible to solve, as people have different skillsets and starting conditions. some people with a bad start manage amazing things, many more fail, and some people with every advantage fail regardless. some people get hopelessly depressed and some just get more manic. a utopia of this sort is impossible.

even when robots do all the work, everyone has a basic level of subsistence that is positively luxurious compared to what we considered subsistence today... but the rich will still own the best property. they will still command the lions share of that robot-generated capital. they will be the first people buying up planets and creating massive dynasties of hyper-capable gene-edited descendants that live forever.

and there will still be people seething about that, in between fucking their sexbot and having a 5-star Michelin meal whipped up by a culinary droid.

>>16689220 (OP)
100% service economy, maybe with some artisanal stuff thrown in. humans will still want to be with other humans, even as the technology gets better, only autists will have 100% buy-in with the machine at first. we're seeing it happen now with chatbots.
probably a whole lot of sex work, until the AI perfects the sex bot.

>>16691362
can confirm, i quit my job three weeks ago, because i hated it.
in hindsight, probably not the best move. but i really felt i had to. and i can afford to be jobless for a little while.
god i wish i enjoyed something that wasn't pure wank. i wish i enjoyed putting up drywall, for example.

>>16691465
Everything in Dune sounds like a shithole, so I sincerely hope we don't go down the path of banning AI like a bunch of luddites.

>>16689656
they shouldn't be too expensive. how much does a new car cost? like 50k say for upper end
a robot probably uses 1/5th the material of a car, so let's call it 5k. maybe 10k because a robot is more sophisticated than a car (a modern car is already incredibly sophisticated).
obviously robots can be produced like cars, it's an assembly line, the same thing getting built over and over to exacting specs. if we can produce tens or hundreds of thousands of cars per year, in a market with suppressed demand (from price inflation) and inflated supply (from car culture), imagine what we can with a new market that has intense demand (from new tech) and no supply (the first robots are not yet a thing).
i suspect the first AI company with an AGI or ASI will buy up factories that are not being used and convert them to robot factories. i imagine shuttered car fabs are perfect for this.
then again maybe the ASI will reveal higher efficiency building new from novel designs, who knows.

>>16692029
whoops i got the math wrong lol i'm a retard
so 10k/20k assuming 1/5th the material
still, i think this is affordable, it's not a house, you could lease this for a few hundred a month. ideally such a robot could do all physical house labor as well as simple errands outside the house
have a live-in servant and a spotless house, worth every penny i believe. even if there is additional energy and maintenance costs.

>>16689220 (OP)
live forever in 1:1 scale VR world as a qt elf girl

>>16691795
essentially because gen X tech bros have realized that immortality is close, but only truly achievable with the great acceleration in scientific progress that AGI will make possible

>>16692086
to be rich is one thing, but to be rich and live forever, that's the dream.
hopefully they'll let high IQ people access the tech even if they're poor fucks. instead of keeping it a secret to avoid rioting.

>>16691803
I used to be all for technology and science when I was in my twenties. I had the mindset that it was better that humans. But then you slowly see that those types of dreams are naive and that in reality it's dystopian. I came to learn that as a human you need other humans, despite how poorly you may function with them. You need another conscious being for what you do to matter.

>>16691824
>>16690490
Would you be satisfied ever only being as good as a minuscule fraction of the AGI? What difference does it even make if you're a tiny bit better than you would have been if you didn't work to better yourself? Whatever you did wouldn't matter to anyone.

Certain roles must be occupied by human beings, such as religious careers. People must also build and maintain these systems. I'm a programmer and understanding things like machine learning and artificial intelligence are important. In fact, programming is for me a means to afford the construction of a masjid and establish a religious career.

We will just do what rich people already do.
Vacations, eating good, sex.

>>16692304
I think that we will have to fight for it

>>16692311
Being rich is relative, people on earth will live better than ever but, the wealthy will love off planet.
O'Neill cylinders, with 10 million humanoid robots per human.
Even the most wealthy kings on earth could not dream of the life they will live.

>>16692465
>>16692311
>>16692304
You all have different ideas presumably based heavily on your own inclinations.
I think what will actually happen could be based a lot on chance and that whatever happens could stick around for a very long time.

>>16692121
The vast majority of people never achieve an olympic gold medal or anything equivalent.

>What difference does it even make if you're a tiny bit better than you would have been if you didn't work to better yourself?

My own personal experience is enhanced by it. Face it, we have short lives, and wasting so much of it degrading ourselves with tedium is far from the ideal life. I don't care if other people or machines are better than I am. I care if I am better than I was yesterday.

>>16692517
You're going to be better then yesterday, then at a point you're going to start being worse than yesterday. What's the point though? I just see it as a life wasted on striving for nothing.

>>16692517
>. I care if I am better than I was yesterday.

No matter what you do or how you live your body fails you.
Embrace the machine!

>>16689220 (OP)
We won't make it that far. The near-future of this species is endless jeets and Africans.

>>16692534
>What's the point though?

Why don't you just kill yourself, right now?

> a life wasted

What isn't a waste of life to you? How is your post not a waste of your life? I ask again, why don't you commit suicide right now, since you're not accomplishing anything other than childish whining on the internet?

I have languages to learn, you pathetic worm.

>>16692538
>No matter what you do or how you live your body fails you.

I look forward to observing what the machines are capable of.

Personally, I seek to find the limits of what my biological machinery are.

>>16692557
>Why don't you just kill yourself, right now?
You didn't answer. What's the point to you?

>why don't you commit suicide right now, since you're not accomplishing anything other than childish whining on the internet?
I'm not whining. I'm asking you questions. This seems to have triggered you.

>you pathetic worm.
Damn. I was being friendly to you.

im defo gonna use AI to summarize this thread

>>16692561
You're a bad chatbot that I vote to be deleted.

>>16692572
Cut your head off, schizo.

>>16689220 (OP)
51% will be paid to monitor the other 49% in case they try to change the situation.

>>16689220 (OP)
We might go back to the era of gentleman scientists, poets, great discoverers and more. Realistically, many will become couch potatoes watching TV.

>>16689220 (OP)
Nothing, we die. Now the question is if the AI stays submissive and it will be the AI company owners culling off the masses of useless eaters, or if AI decides to cull off even its masters.

>>16692517
The problem with being worse then AGI is that AI will inevitably be cheaper. Right now its really expensive because of our unoptimized hardwere, but the underlying basics are in favor of AIs. Brain requires sugars to power the brain, which need to be pumped into the brain, then they are broken through respiration into energy which is then used to synthesize ATP, and then that ATP is delivered alongside the axons of the nerve cells to pump protons from one side to the other so that when the electric signal comes, they can be moved from one side of membrane to the other. I don't think you need to know why this whole process is far more inefficient then just simply making electricity and running it through a transistor which is so small it is starting to have problems on the atomic scale. So you will inevitably be replaced by it because of your inefficiency.

>>16693549
>that AI will inevitably be cheaper.

Your entire view of the world is based on monopoly funny money. Fiat currency is just bits in a memory drive. You talk about efficiency without any reference to what end you are optimizing for.

You're a genetic slave, doomed for obsolescence.

>>16693559
AGI by definition should be able to do everything human can do. In terms of efficiency AI is more efficient in every aspect, be it energy cost, money cost, carbon cost, whatever. To run a neuron like I said you need somewhat balanced diet, then support the whole human body because humans aren't just brains, and then you also need bunch of recreational stuff and ways to make you not go insane, so even in the most cruel most optimized conditions AI could mog humans enough to make eradicating humanity its main priority in almost any goal there is.

A few things of note:

1. There's no fixed amount of labor to be performed in the world. We can always do more because we're moved by envy, not greed. The endgame would either be a garden earth, the stars or simulators. In either case, it's fine.

2. The only losers in automation are the economic zones that don't automate. See Switzerland, a country famed by the highest quality manufacturing, highly automated, and Malawi, likely the worst shithole on earth, with neolithic agricultural practices. Which one has more unemployment?

3. The transition to highly robotic manual labor will be akin to the transition from farming-dominant societies to industrial societies, but much less drastic, because manufacturing already accounts for a minority of employment. It won't cause any upheavals.

3. The endpoint of the economy is family spending and government spending. They will have to keep up somehow or supply will simply disappear. "
>Oh, no one can afford slipshakes on zogdonalds anymore!!!!
Then zogdonalds will fail.
But there's zero chance that a massive uptake in product will lead to poverty. Again, look at Switzerland, Norway.
The only bad piece on this table is land prices. Governments will likely have to be more agressive in avoiding real estate price inflation in the future.

>>16693549
>The problem with being worse then AGI is that AI will inevitably be cheaper.
Not actually inevitable. The current major providers are operating at obscene losses over the last year and a half to push mass adoption. They're kept afloat more by influx of investment than by actual userbases, and basically none of the industry as it currently stands is sustainable.

If AGI is as costly or more costly to run than current LLMs, there's a good chance that the economics of the labor (or compute, if you want to frame it that way) you get from a cheap employee are still superior. See: hiring Indians to pretend to be an AI. Might be part of why so much pro-AI stuff seems to come from there; they're sick of being Mechanical Turks and they're inundated with corporate propaganda to keep them from realizing they're training the machines that will destroy their current livelihoods.

It cannot be stressed enough just how energetically inefficient compared to fleshbags ALL computing is; your brain uses less energy under maximum load than the average desktop PC by a factor of 20-50, and even many mentally challenged humans are more autonomous and more flexible than such a device. (There's a common misconception that learning disabilities prevent learning or problem solving; most of the time what they really do is make learning and problem solving slower. There isn't actually an IQ-derived ceiling, there's just not an impetus in the education industry to spend the requisite time and effort on these people to get them to parity levels of knowledge.)

As an aside, Nvidia's strategy seems to be to just dump more power into denser silicon for performance gains, so at scale it'll be really interesting to see if AMD's apparent push into more efficient ASICs will pay off - we've already seen something similar annihilate the market for GPUs for crypto mining. Watts per ANN node is going to be a very important factor for big players moving forward.

>>16693577
Two things.
One AI is expensive because people use it for slop images and videos or they ask ChatGPT 50 times and generate almost a book worth of words just to write one page of content. But even now the cases where AI can do the full job of a human AI is 1000x cheaper, and that will remain even if you remove the loss leader subsidy.
Two, AI programs, especially LLMs are incredibly unoptimized and inefficient, they reread 5000 words every time they write a new word and run on computers designed for programming, browsing internet and playing videogames. The transformer architecture is incredibly inefficient and has been optimized recently by Chinese with DeepSeek that supposedly only costed 5 million to train. Human brains run much slower then LLMs and your brain could probably not run a videogame like desktop PC. We just have more energy conserving brain architecture that does not light up all neurons at the same time every microsecond, which is how we get the factor of 20 lower energy cost. Human brains also like I said run on balanced diet, so not just sugar and water, but also amino acids, vitamins, minerals and so on, and those when converted to calories are much more expensive then simple electricity that also has million different way of being generated.

>>16692311
Do you need to fight for food or water today?
In the past they did, now in developed nations such basic stuff is avaliable to anyone.
AI is just another step in the same direction.

>>16692034
Robots will be sold to companies as a service.

>>16693585
This is just a market imbalance, then.
The reason so much compute is being allocated to openai and google is that their total addressable markets is literally everyone on earth except China and this is like crack to financiers.
Energy won't be a problem for much longer, anyway.

>>16693574
Again, you don't actually say what the goal is. You're arguing for humans to go extinct to optimize the production of paperclips.

Lousy.

I'm saying we should be optimizing the experience of life for all things on Earth, and that could include living machines.

Even sentient machines without purpose will suffer the void of meaning.

>>16693585
>Human brains run much slower then LLMs
This is not the case, and it isn't close.

The rate limiter to human text output is actually typing speed, and anguage production is localized to a pretty specific area of the brain on one side, Broca's area.

The rate limiter to producing visual art is likewise more a physical expression limit. So long as you don't have aphantasia, you should be able to visualize things at a speed and efficiency well exceeding the current generative tech, and odds are what you imagine is continuous and maybe even multisensory. And it's also exactly accurate to what you imagine - you aren't rolling and re-rolling because you imagined a horse when you wanted to imagine a dragon, and you can make changes to details on the fly.

I still wonder how much AI art evangelism is from people with aphantasia. Exercise your imagination, it's good for you.

Regardless, much of your brain is dedicated to sensory processing, and in visual processing especially your brain still blows perceptrons out of the water in terms of speed and efficiency. Even where a perceptron may categorize a given image slightly faster, it can't categorize a continuous stream of visual information without obscene power draw and compute scale. And it can't actively learn while doing so.

We're talking, even with current tech, a 10,000+ watt system the size of a small building approximating a small portion of what your brain does for 20 watts. The inefficiency of converting energy to consumable calories is so far off the magnitude scale we're talking about it's irrelevant to the comparison.

>has been optimized
Well... no. These inefficiencies come from randomness and other fundamental architectural limitations. They never fully go away for LLMs because the systems ARE stochastic no matter how much tech bros kvetch about it, and a monetary cost claim for training efficiency isn't relevant to operation efficiency because training is a one-time event per model version.

AGI is a meme, all attempts at producing a 200 iq robot will fail
In a few years the hype will die down
It will boost labor productivity but never totally replace human labor
Robots and AI need human oversight for safety reasons
But they boost productivity by 2-10x
Most future jobs will be like playing videogames
All the self driving car companies have humans overseeing them from remote platforms
The only proper definition of an "AGI" is a group of humans using robots

>>16693670
Wrong. AI is superior to humans. Face it. You lost. Just accept that AI will replace you.

>>16693738
Look, I know this is bait, but this sentiment held earnestly because of personal investment is the exact sort of thing that pisses me off in the field. We COULD be moving towards superhuman AGI, but sunk cost fallacy and tech ignorance has turned discussion around dynamically weighted dice into a religion.

The loudest voices about it online want the things they've been told about LLMs by marketing campaigns they couldn't distinguish from genuine interaction to be real. Meanwhile I want actual intelligence on artificial hardware, which we do not have and are not approaching (the best we can do is approximate the probability distribution across a STATIC set of outputs from real intelligence).

We've already seen AI research stagnate for decades after chasing a dead end (this exact one, BTW - using perceptrons for everything; we just didn't have the hardware to go ham on them before) before, and I'm not looking forward to the fallout of an even larger bubble popping in the space.

We will die out and they'll continue developing up untill the moment they get so advanced that they're practically indistinguishable from homo sapiens. They themselves won't know that they once been slave machines and fuck, it might've happen already

>>16693761
The bubble will pop but there wont be AI winter ever again. The first bubble popped because AI proved to be useless dogshit. Now AI is used for many applications that aren't obscure at all. So the worst thing that could happen to it is lowered investment in giant wasteful models, but still equal amount used for actual research of new methods.

>>16693735
>It will boost labor productivity but never totally replace human labor
Sure? First line helpdesk is being replaced with LLM already.

>>16693735
I find it as quite the opposite. AI is so fucking dumb and hard to work with that it can only work as a replacement, because otherwise it just spits junk at you.

>>16693769
There's no such thing as a proper bubble because there are only a few already exceptionally profitable companies who are really benefiting from the hype.
Nvidia might be overvalued (though not really, if you take their price/earnings number) but their hardware and software has many useful applications already. They won't go bust.
Microsoft is still massively profitable and cloud alone will carry them forever.
Google too. Their ad revenue is growing and will likely keep doing so forever.
There's no bubble except for venture capitalists investing in retarded AI startups.
OpenAI might eventually crumble under its own weight, but their actual products are already fine and on track to be profitable. They won't be doing seasonal mega traning runs anymore.
I'm sorry, anon, but even if we never get to AGI, no massive crash is going to happen.

>>16693775
That's about the only industry use case for it. Funny thing is it's not new automation - we've had automated phone systems for decades, for example.

>>16689220 (OP)
Get drafted to fight robots. Drafting a human is cheaper than building a robot.

>>16694493
The life of a human can't compare with the perfection of the digital

>>16689220 (OP)
whatever the fuck they want tf kind of question is that

>>16695119
The elites will have some control over the machines’ programming I think.
Why would they play fair all of a sudden?

>>16695145
because none of this matters and fair games are more fun

>>16689220 (OP)
Starve, according to the malthusian misanthrope collectivists. Are you stupid?

>>16689220 (OP)
I’ll fix the robots.
Mechanics are gods.

>>16689714
>Humans will probably lease out their brain power if human brains even amount to any significant computing power
Yea I saw that episode in the new season of black mirror too.

I am learning how to build advanced machine learning models. It is frying my brain, but am rapidly learning a lot of new math. Things get difficult when you are studying algorithms invented in the last 10-20 years.

>>16697569
I credit Islam for providing the soberness and discipline I need to apply my mind. I am sober 24/7 and motivated by a need to help my people. Lots of tea and a pomodoro timer between daily prayers. Being sober all the time and maintaining a stable mood is a severe advantage in life.