So, you are saying, I should mix my cocaine with twix bars for maximum efficiency? (Would still be stupid, but now more efficiently)??
post
See, the thing is, I watch piss porn. Hear me out. I told my friend that the thing is, to do piss porn, you kind of have to be into it. You could try and fake it, but it wouldn't be very convincing. So, my contention is, piss porn is more genuine than other types of porn, because the people partaking are statistically more likely to enjoy doing that type of porn. Which is great, I think, because then they really get into it, which is hot. It's that enjoyment that gets me off. Their enjoyment.
She said, "Krooklochurm, you're an idiot. Anyone can fake liking getting pissed in the face."
So I said, "Well, if you're so adamant, get in the tub and I'll piss in your mouth, and let's see if it's as easy as you claim."
So she said, "All right. If I can fist you in the ass afterwards."
Which I felt was a fair deal, so I took it.
My (formal) position was strengthened significantly by the former event. And I can also attest that I could not convincingly fake enjoying being ass-fisted.
What does that have to do with anything, you ask? Genuinity. The real deal. That's what.
What the fuck did I just read
Plot twist, she really did like getting pissed on, and she knew it ahead of time. She was gaming you for that golden shower.
Honestly disgusted to read that, but you do you...:) If you feel like thats your thing, I guess thats what you are here to enjoy.
I could see it. A lot of people dont like it, and its not my personal thing, but sex can get messy so its p much whatever. Always makes me think of this though:

This is nice to be confused with shit porn. Which is just not very good.
Wasn't there an article posted yesterday about a group trying to create a biological computer that was living cells do to their efficiency of use on less power? (They are far from close, they basically took skin cells, ionized them, and had no idea how they were going to get them to stay alive long term yet.
Even that won't be anywhere close to the efficiency of neurons.
And actual neurons are not comparable to transistors at all. For starters the behaviour is completely different, closer to more complex logic gates built from transistors, and they're multi-pathway, AND don't behave as binary as transistors do.
Which is why AI technology needs so much power. We're basically virtualising a badly understood version of our own brains. Think of it like, say, PlayStation 4 emulation - it's kinda working but most details are unknown and therefore don't work well, or at best have a "close enough" approximaion of behaviour, at the cost of more resource usage. And virtualisation will always be costly.
Or, I guess, a better example would be one of the many currently trending translation layers (e.g. SteamOS's Proton or macOS' Rosetta or whatever Microsoft was cooking for Windows for the same purpose, but also kinda FEX and Box86/Box64), versus virtual machines. The latter being an approximation of how AI relates to our brains (and by AI here I mean neural network based AI applications, not just LLMs).
I'm very much aware of FPGA-style attempts, however I do feel the need to point out that FPGAs (and FPGA style computing) is even more hardware-strained than emulation.
For example, current mainstream emulation FPGA DE10 Nano has up to 110k LE/LUT, and that gets you just barely passable PS1 emulation (primarily, it's great for GBA emu, and mid to late 80s, early 90s game console hardware emulation). In fact it's not even as performant as GBA emulation on ARM - it uses more power, costs more, and the only benefit is true to OG hardware execution (which isn't always true for emulation).
Simply said, while FPGAs provide versatility, they're also much less performant than similarly priced SoCs with emulation of the specific architecture.
I think the entire idea of ai and the Internet in general taking up power and water needs to be fleshed out and explained to everyone. Even to me it’s a vague notion, I heard about it a few years back but can’t explain it to someone like my parents who would have no idea the Internet requires water to run
It's not too hard. AI requires a LOT of work. Work requires energy. Some energy is wasted during this and the byproduct is heat. The heat has to be removed for many reasons, and water is very good at doing that.
It's like sweating, it cools you down. But you need water to sweat.
it's really good at writing termination notices without making middle managers feel bad about letting their employees go.
I think we're at a point were the hardware right now does not fit with the algorithm being used. Since they take so much power due to our computers being digital. Having a transistor only capable of holding 1 state (0V or 5V usually) is inefficient. The heat add up as you multiply especially with LLMs. There seems to be a potential for analog where a transistor acts more on a range 0 - 5v. Which in theory could store more information or directly represent what LLM runs on (floating point). For more context 1 float tends to be 32bits. 1 bit is 1 transistor so 1 float = 32 transistor. While an analog transistor could be 1 float = 1 analog transistor.
My mistake, I made it look like that's a fact I'll edit that as my opinion. Though here is one I can find.
https://www.nature.com/articles/s43588-024-00753-x
Article also https://spectrum.ieee.org/analog-ai-2669898661
We use about 20% of our caloric intake (at rest, not doing math) for our bio intelligence. Having superpowers of social organization is expensive and power hungry.
So it's really no surprise that the computation machines that can run AI require tens of megawatts to think.
I'd trade cocaine for massive amounts of caffeine!
How much have you got? I've got about 3kg of coffee.
It's not artifical intelligence. A Large Language Model is not intelligent.
And yes yes, scientifically, LLM belongs there and whatnot. But important is, what the people expect.
I completely agree that LLMs aren't intelligent. On the other hand, I'm not sure most of what we call intelligence in human behavior is any more intelligent than what LLMs do.
We are certainly capable of a class of intelligence that LLMs can't even approach, but most of us aren't using it most of the time. Even much (not all) of our boundary pushing science is just iterating algorithms that made the last discoveries.
Not to be pedantic, but the original use of the word intelligence in this context was “gathered digested information.”
Unfortunately, during the VC funding rounds for this, “intelligence” became the “thinky meat brain” type, and a marketing term associated with personhood, and the intense personalization along with it.
All it really takes is the Limitless pill. The protagonist in the movie Limitless gets so much done after taking the Limitless pill in the movie Limitless. Limitless is a movie about a man who discovers the Limitless pill to help him accomplish almost limitless amount of things. But he learns that there is a limit that the Limitless pill can do if he takes it for too long. Because the Limitless pill isn’t really limitless.
He learns the hard way that he needs to find a way to limit his dependency on the Limitless pill.
It gets somewhat retconned in the TV series though where he's basically just this political genius who secretly runs everything.
The TV show is a lot better than it has any right to be, it's a shame it only got one season.
Ya evolution is pretty cool.
And on that note the human physiology sacrifices quite a bit for its intelligence.
For example the reason humans come out as babies is because if they came out with a full sized brains they'd kill the mother!
It's all about the most proficient use of energy.
Back when I was in college, grad students in physics seemed to live on frozen burritos (thawed and cooked in the microwave), and cherry coke.
It's an efficient, if somewhat finicky intelligence. It checks out commander.
yeah i mean ofc if you also put everyone in the world that that datacentre is serving in a human datacentre, I'm sure it'd also consume tons of power (in food)
1 gram of cocain equals roughly 150 grams of CO2 emissions due to production and shipping etc, plus the effect wears of very quickly. Cocain Als destroys your nostrils, it's really, really bad. I would advice amphetamine instead. Can also be taken orally, for instance in the medicinal form of dexamphetamine. Another side effect is that you aren't hungry anymore so you don't need the Twix. Just dexamphetamine and you are able to achieve your goals better like becoming a dictator (like Hitler, he got daily shots of amphetamine) or invade France if you want (the German army had amphetamine pills which helped them advance into France day and night. The French assumed they would stop during the night to rest but since they didn't, the French greatly miscalculated and were completely overrun. Thats why you should use amphetamine kids). It also really helps with ADHD to focus on things and think clearly.
At this point, I feel like they actually excel at classifying people by political views and all the red number as covered by spy agencies... call it a conspiracy, but its my shot in the dark
Yeah, Adderall sounds terrible. Most medications are terrible. I found the only medication’s that I needed is exercise and having a good friend everything else just makes you numb and the worst thing. Yeah limitless might be Adderall terrible stuff. Probably people get hooked on stuff and now you don’t even know what’s in that stuff anymore but you know the worst thing was going to a psychologist office and he’s asking you what pills you want I guess it hit the doctor. I never went back to a psychologist ever again and everyone wants me to cure myself with my numbing drugs. I tried some. I tried to so many none of them. They ruined my concentration and they make me dull. There’s no love in life on pills. I know a couple people who abuse prescription drugs, just as bad as anything just as bad as alcohol just as bad as anything out there I signed off of weed years ago cause I just saw how devastating it is and it doesn’t even cure cancer. It’s just people using an excuse to escape. I’m tired of escaping and just I just want fresh air exercise, warm heat and my best friend that’s all I need in life.
Compare 1 human to one LLM session, instead of one human to all LLMs on Earth, and you’ll see that we’re way less efficient
Also, one human brain training takes feeding for years before it can do useful things.
Does that actually add up, though?
Google released stats recently that the median Gemini prompt consumes about 0.24 watt hours of electricity.
For humans performing knowledge based labor, how many prompts is that worth per hour? Let's say that the average knowledge worker is about as productive as one good prompt every 5 minutes, so 12 per hour or 96 per 8-hour workday.
Let's also generously assume that about 25% of the prompts' output are actually useful, and that the median is actually close to the mean (in real life, I would expect both to be significantly worse for the LLM, but let's go with those assumptions for now).
So on the one hand, we have a machine doing 384 prompts (75% of which are discarded), for 92 watt hours of energy, which works out to be 80 kilocalories.
On the other hand, we have a human doing 8 hours of knowledge work, probably burning about 500 calories worth of energy during that sedentary shift.
You can probably see that the specific tasks can be worked through so that some classes of workers might be worth many, many LLM prompts, and some people might be worth more or less energy.
But if averages are within an order of magnitude, we should see that plenty of people are still more energy efficient than the computers. And plenty aren't.
You completely forgot that workers only work one time the shift per day. If you account for the total energy required to do a project, and assume that a human would do alone X in 5 days, then wouldn’t it be better to use prompting as well, which would theoretically, in this model, make X feasible in 2.5 days? Sure, the non-work calories consumption of a human is inevitable, but when strictly talking about productivity, you can make an individual be 2x more productive for a lot less than their daily calories consumption
I however appreciate your analysis and the time you took to write this. It’s really nice to see a comparison. I would have imagined it consumed a lot less than that.
If you compare this to letting a desktop computer run though… it’s negligible. I know mine uses about 80W when idle
You need a nuclear power plant not for a single AI, but for several million instances of it.
Don’t forget that you can run full OSS ChatGPT on a single Mac Mini.
AI is sick, and only silly people can deny that. Yes, future is little scary, but scary things should not stop us from progress.
Yes, we do have a lot of AI slop, but don’t forget that AI can be an exceptionally good tool in right hands, and its spheres of usage grow every month. I can’t wait for generative tools for gauss splattering things, for example - because it can lead games developing process on another level.
Isn't it more like they're comparing all the hamburgers and everything else you have eating since you were born?
That's what they're doing with AI enegry usages isn't it? I thought it was including the training which is where the greatest costs come from vs just daily running.
all 70 comments