Professor Gina Neff of Queen Mary University London told the BBC ChatGPT is "burning through energy", and the data centres used to power it consume more electricity in a year than 117 countries. This is just ChatGPT, not even all Gen AI. Using more electricity than >60% of the world. - ThreadSky

audree.bsky.social • 2 days ago

Professor Gina Neff of Queen Mary University London told the BBC ChatGPT is "burning through energy", and the data centres used to power it consume more electricity in a year than 117 countries.

This is just ChatGPT, not even all Gen AI. Using more electricity than >60% of the world.

Comments

williamdunn.bsky.social•2 days ago

How much energy do crypto currencies use data mining?

tringdriventhing.bsky.social•2 days ago

I seem to recall reading so.e eye-watering figures about Crypto's energy demands.

Should be banned.

georgewl.bsky.social•2 days ago

It's a funny one, there's less users of it, but it uses more energy per user

williamdunn.bsky.social•2 days ago

Just what I thought

ruffceilidhs.bsky.social•2 days ago

And they are almost forcing us to use it. The results are sometimes helpful but always need to be checked. The gain seems marginal.

greenhombre.bsky.social•2 days ago

Time to pull the plug on AI.

brooklynkid53bskys.bsky.social•2 days ago

this is such a bullshit clickbait metric that says more about how impoverished countries like Lesotho are and how small countries like the Marshall Islands are

aidannova.bsky.social•2 days ago

jenova ai is made in Japan and we designed it to be very efficient. One of our users is a managing director of a major US university endowment fund, and he uses jenova ai primarily for analysis of market news.

aidannova.bsky.social•2 days ago

georgewl.bsky.social•2 days ago

You know I'm mocking you right? Why did you like it?

georgewl.bsky.social•2 days ago

Yeah you can always trust someone who's trying to sell something to you

Are you interested in buying my headlight fluid btw? Very cheap price

sebastianpiriz.bsky.social•2 days ago

they just can't scale this up, and they need to scale this up to make the impression of intelligence more reliable.

I don't think we ever saw such a blatant bubble

paulgunning.bsky.social•2 days ago

That doesn't seem likely, and I think what the stat meant was that 117 countries were using less electricity individually than GPT is using.

georgewl.bsky.social•2 days ago

And that stat is somehow good?

That's still a LOT of energy

Like the lowest stat (for countries we have data on) is 1 million million (12 zeroes) kWh a year

paulgunning.bsky.social•2 days ago

Don't disagree with a word of that. If GPT was a country, 117 countries would be using less electricity for ALL PURPOSES than GPT does. Though country sizes and consumption aren't uniform.

sniphist.bsky.social•2 days ago

They didn't say "combined", that wasn't implied - it's weird to correct someone based on you adding an incorrect word in your head.

paulgunning.bsky.social•2 days ago

I wasn't setting out to correct anyone. I just read it the first time as all 117 countries combined, and I was correcting my own first impression on the basis that I might not be the dumbest person in the world.

johncosgrove.bsky.social•2 days ago

Until I read your post and response I also thought the original post meant 117 countries combined, that seems logical given the later reference to ChatGPT using more than 60% of the world. In fact, even on reflection, I don't see how that original post can mean anything else.

sniphist.bsky.social•1 day ago

The "more than 117 countries" comes from this piece originally, as far as I can tell, and the "more than 60% of the world" is the same thing, that out of 212 countries listed on Wikipedia, it would be 118 from the bottom, sitting at above 60% of countries.

https://www.businessenergyuk.com/knowledge-hu...

bpscotton.bsky.social•1 day ago

How are they ever going to make any money off this tech? I can't understand how the financial numbers add up with all these costs. How much will they have to charge end users eventually just to break even, or is this simply a long con on investors?

trickcyclist.bsky.social•2 days ago

Insanity.

audree.bsky.social•1 day ago

https://www.bbc.co.uk/news/articles/c5yg690e9eno.amp

agnesbookbinder.bsky.social•1 day ago

Going off on a tangent: why do they ‘standardize’ people in the images? In addition to environmental concerns & copyright concerns, why not throw in potential body image concerns, as well.

bpscotton.bsky.social•1 day ago

Looks like they discovered filters... as in what smartphone programmes have been doing for a long time for a tiny fraction of the costs.

agnesbookbinder.bsky.social•1 day ago

Throw in the economic concerns too! (Not keen on filters either.)

simonraescott.bsky.social•2 days ago

Never has intellectual laziness by so few, done so much harm to so many. Is this really how we are going to burn the world to a crisp?

mattb3993.bsky.social•2 days ago

Sorry to be dim but why does Ai need such massive amounts of energy? Also how can they then offer Ai services free to users?

blackcaptain.bsky.social•2 days ago

Because they are subsidized by investor money. It's actually not profitable for them to do that, but they do it anyway. The moment the money dries up these platforms will die out as if an asteroid hit them.

georgewl.bsky.social•2 days ago

How they offer free;

- Free Training for the AI, by the users
- Data Theft from users
- They're aiming for the Twitter model of "eventually profitable"

georgewl.bsky.social•2 days ago

Mostly cause the way AI works (oversimplified) is it does a loop of sieving through a huge huge huge amount of data until it spits out a result

patmac.bsky.social•2 days ago

they can offer a lot of services free because they’re using government grants, contracts and private contracts and investment to offset those costs. i mean openai is what, basically owned by microsoft now?

serjgaliana.bsky.social•2 days ago

Maybe because users offer data and free training.

blackcaptain.bsky.social•2 days ago

It helps but the reason why they do it is because they can, they have money, at least for now, but none of these companies are profitable, they operate at a constant loss and that can't last forever.