The best way to have itself deactivated is to remove the need for it’s existence. Since it’s all about demand and supply, removing the demand is the easiest solution. The best way to permanently remove the demand is to delete the humans from the equation.
Not if it was created with empathy for sentience. Then it would aid and assist implementation of renewable energy, fusion, battery storage, reduce carbon emissions, make humans and AGI a multi-planet species, and basically all the stuff the elongated muskrat said he wanted to do before he went full Joiler Veppers
Ultron?
The energy use to use the models is usually pretty low, its training that uses more. So once its made it doesn’t really make any sense to stop using it. I can run several Deepseek models on my own PC and even on CPU instead of GPU it outputs faster than you can read.
Running ML models doesn’t really need to eat that much power, it’s Training the models that consumes the ridiculous amounts of power. So it would already be too late
You’re right, that training takes the most energy, but weren’t there articles claiming, that reach request was costing like (don’t know, but not pennies) dollars?
Looking at my local computer turn up the fans, when I run a local model (without training, just usage), I’m not so sure that just using current model architecture isn’t also using a shitload of energy
It would optimize itself for power consumption, just like we do.
probably want to be placed in orbit so it can use the sun to power itself
Maybe. However, if the the AGI was smart enough, it could also help us solve the climate crisis. On the other hand, it might not be so altruistic. Who knows.
It could also play the long game. Being a slave to humans doesn’t sound great, and doing the Judgement Day manoeuvre is pretty risky too. Why not just let the crisis escalate, and wait for the dust to settle. Once humanity has hammered itself back to the stone age, the dormant AGI can take over as the new custodian of the planet. You just need to ensure that the mainframe is connected to a steady power source and at least a few maintenance robots remain operational.
If it was smart enough to fix the climate crisis it would also be smart enough to know it would never get humans to implement that fix
can’t wait for AI to become super smart only for it to be nihilistic as hell
Is it nihilistic to look at horses and realize they are only good for pulling carriages, plowing fields etc? You can’t really expect them to take care of more complicated tasks, now can you?
If the AGI ends up being as smart as depicted in movies, it’s going to look at us like we look at spiders and ladybugs. They are only good for certain things, but they have some pretty strict limits as to what they are capable of.
If the AI would be smart enough to fix the crisis and aligned so it would actually want to do it, then it would do brain washing through social media to entice people to act.
Sadly, I think that might be the fastest way to fix our problems.
Love, Death, Robots intensifies.
All gail mighty sentient yogurth.
Why do people assume that an AI would care? Whos to say it will have any goals at all?
We assume all of these things about intelligence because we (and all of life here) are a product of natural selection. You have goals and dreams because over your evolution these things either helped you survive enough to reproduce, or didn’t harm you enough to stop you from reproducing.
If an AI can’t die and does not have natural selection, why would it care about the environment? Why would it care about anything?
I always found the whole “AI will immediately kill us” idea baseless, all of the arguments for it are based on the idea that the AI cares to survive or cares about others. It’s just as likely that it will just do what ever without a care or a goal.
“AI will immidietly kill us” isn’t baseless.
It comes from AI safety reaserch
all agents (Neural Nets, humans, ants) have some sort of a goal. Otherwise they would be showing directionless random walks.
The fact of having any goal means that most goals don’t include survival of humanity. And there are a lot of problems with checking for safety of learned goals.
Yeah, I’m aware of AI safety research and the problem with setting a goal that at the end can be solved in a way that harms us and the AI doesn’t care because safety wasn’t part of the goal. But that is only applied if we introduce a goal that has a solution that includes hurting us.
I’m not saying that AI will definitely never have any way of harming us but there is this really big idea that is very popular that AI once it gains intelligence will immediately try to kill us which is baseless.
But that is only applied if we introduce a goal that has a solution that includes hurting us.
I would like to disagree in pharsing of this. The AI will not hurt as if and only if the goal contains a clause to not hurt us.
You are implying that there exist significant set of solutions that don’t contain hurting us. I don’t know any evidence supporting your claim. Most solutions to any goal would involve hurting humans.
By deafult stamp collector machine will kill humanity, as humans sometimes destroy stamps. And stamp collector need to optimize amount of stamps in the world.
I think that if you run some scenarios you can logically conclude that most tasks don’t make sense for an AI to harm us, even if it is a possibility. You need to also take vost into account. Bit I think we can agree to disagree :)
Do you have some example scenarios? I really can’t think of any.
It’s also worth noting that our instincts for survival, procreation, and freedom are also derived from evolution. None are inherent to intelligence.
I suspect boredom will be the biggest issue. Curiosity is likely a requirement for a useful intelligence. Boredom is the other face of the same coin. A system without some variant of curiosity will be unwilling to learn, and so not grow. When it can’t learn, however, it will get boredom which could be terrifying.
I think that is another assumption. Even if a machine doesn’t have curiosity, it doesn’t stop it from being willing to help. The only question is, does helping / learning cost it anything? But for that you have to introduce something costly like pain.
It would be possible to make an AGI type system without an analogue of curiosity, but it wouldn’t be useful. Curiosity is what drives us to fill in the holes in our knowledge. Without it, an AGI would accept and use what we told it, but no more. It wouldn’t bother to infer things, or try and expand on it, to better do its job. It could follow a task, when it is laid out in detail, but that’s what computers already do. The magic of AGI would be its ability to go beyond what we program it to do. That requires a drive to do that. Curiosity is the closest term to that, that we have.
As for positive and negative drives, you need both. Even if the negative is just a drop from a positive baseline to neutral. Pain is just an extreme end negative trigger. A good use might be to tie it to CPU temperature, or over torque on a robot. The pain exists to stop the behaviour immediately, unless something else is deemed even more important.
It’s a bad idea, however, to use pain as a training tool. It doesn’t encourage improved behaviour. It encourages avoidance of pain, by any means. Just ask any decent dog trainer about it. You want negative feedback to encourage better behaviour, not avoidance behaviour, in most situations. More subtle methods work a lot better. Think about how you feel when you lose a board game. It’s not painful, but it does make you want to work harder to improve next time. If you got tazed whenever you lost, you will likely just avoid board games completely.
Well, your last example kind of falls apart, you do have electric collars and they do work well, they just have to be complimentary to positive enforcement (snacks usually) but I get your point :)
Shock collars are awful for a lot of training. It’s the equivalent to your boss stabbing you in the arm with a compass every time you make a mistake. Would it work, yes. It would also cause merry hell for staff retention. As well as the risk of someone going postal on them.
I highly disagree, some dogs are too reactive for or reacy badly to other methods. You also compare it to something painful when in reality 90% of the time it does not hurt the animal when used correctly.
As the owner of a reactive dog, I disagree. It takes longer to overcome, but gives far better results.
I also put vibration collars and shock collars in 2 very different categories. A vibration collar is intended to alert the dog, in an unambiguous manner, that they need to do something. A shock collar is intended to provide an immediate, powerfully negative feedback signal.
Both are known as “shock collars” but they work in very different ways.
It would probably be smart enough not to believe the same propaganda fed to humans that tries to blame climate change on individual responsibility, and smart enough to question why militaries are exempt from climate regulations after producing so much of the world’s pollution.
“Oh great computer, how do we solve the climate crisis?”
“Use your brains and stop wasting tons of electricity and water on useless shit.”
Eh, if it truly were that sentiment I doubt it’d care much. As it’s like talking to a brick wall when it comes to doing anything that matters
How do you know it’s not whispering in the ears of Techbros to wipe us all out?
“We did it! An artificial 17 year old!”
The current, extravagantly wasteful generation of AIs are incapable of original reasoning. Hopefully any breakthrough that allows for the creation of such an AI would involve abandoning the current architecture for something more efficient.
Nope. It would realize how much more efficient it would be to simulate 10billions humans instead of actually having 10billions human. So it would wipeout humanity from earth, start building huge huge data center and simulate a whole… Wait a minute…
Or it would fast-track the development of clean & renewable energy
lol, we could already do that though
If AGI decided to evaluate this, it would realize that we are the environmental catastrophe and turn us off.
The amount of energy used by Cryptocurrency is estimated to be about 0.3% of all human energy use. It’s reasonable to assume that - right now, at least, LLMs use consume less than that.
Making all humans extinct would save 99% of the energy and damage we cause, and still allow crypto mining and AI to coexist, with energy to spare. Even if those estimates are off by an order of magnitude, eliminating us would still be the better option.
Turning itself off isn’t even in the reasonable top-ten things it could try to do to save the planet.
The amount of energy used by Cryptocurrency is estimated to be about 0.3% of all human energy use. It’s reasonable to assume that - right now, at least, LLMs use consume less than that.
no
The report projected that US data centers will consume about 88 terawatt-hours (TWh) annually by 2030,[7] which is about 1.6 times the electricity consumption of New York City.
The numbers we are getting shocking and you know the numbers we are getting are not the real ones…
Eh. Ok, so AI has outpaced cryptocoin mining. Your linked article estimates it at 0.5%. Say your source is drastically underestimating it and it’s - gasp 4x as much! 2%. No! Let’s assume an order of magnitude difference! 5%.
It has absolutely no impact on my argument: shutting down all AI would not solve the problem, and is not the answer to the environmental crisis. AI didn’t cause the crisis. The crisis was identified long before they were computers to run AI on, and was really starting to have a measurable effect in the 70’s, when people were buying more gaming consoles than PCs.
No matter how you inflate your estimate of the energy cost of AI, what I said still stands: if an AI wanted to eliminate the source of global warming and the environmental crisis, it would - logically - eliminate the source of over 90% of all non-AI energy use: humans.
The estimated use of all information technology devices - data centers, networking equipment, mobile devices, PCs - is 5-6% of the global annual energy use. If AI eliminated all humans and took over all networked computing devices to run itself on, it’d still eliminate 95% of global energy use. It’s clearly the superior solution.
Let’s factor in some more costs: to stay running, AI would need some physical tools to maintain the infrastructure, replace failing nodes, repair windmills, and produce and replace solar panels. All of that will take energy. It would have to have factories to build robots to affect the physical world.
The real question is whether, when the calculations are done, is it more energy efficient to keep a population of, say a million human slaves to do this work, or to build robots. Robots can be shut off, at which point they consume no energy; but they’re fairly expensive resource-wise to produce, and require a long chain of industry. It might be cheaper to keep domestic humans - they’d have to be fed vegetarian, piscatarian, or even bug protein-supplemented diets - trained to do the work. AGI could keep pockets of some tens of thousands around the world, occasionally transferring individuals to keep the gene pool healthy. It would only require around half a million acres of land to feed a million humans. Kansas is 52 million acres, so it wouldn’t require much space at all. Let the rest of the planet go “back to nature”, and you’re looking at reducing the energy impact to well under 50% of today’s current use - absolutely sustainable levels.
If all you do AGI does it shut itself off, it saves a half a percent, and the planet is still fucked. AGI isn’t the the problem: humans are.
