This is the Guardian.

Artificial intelligence companies have some pretty lofty ambitions for what the technology

could achieve, from curing diseases to eliminating poverty.

But there's a snag, one that could bring it all crashing back down to earth.

AIs thirst for energy.

Take a look at Google's latest environmental report. It revealed that its greenhouse gas emissions,

generally 50% in the past five years on the back of its expansion of data centers that power

AI systems.

Microsoft is also having to rethink its climate goals in the light of its AI push. But proponents like Bill Gates argue, the investment is worth it, because the tech could also be the key to solving the climate crisis.

We're here showing the state of progress and, you know, makes me optimistic that these

technologies will get out and start to bring emissions down. So we're asking, how big is AI's energy problem? Can an industry whose emissions are

going up help the rest of the worlds to go down.

And, will it happen in time?

From the Guardian, I'm Madeline Finley,

and this is Science Weekly.

To unravel all these questions, today I'm joined by the Guardians UK technology editor Alex Hearn and

Energy Correspondent Jillian Ambrose. Alex I'm going to turn to you first. I think the most recognizable place people will have interacted with

AI these days is chat gPT so I'm going to take that as our example if If I ask chat gp-p-tte to write me a speech about mushrooms,

step me through where all the energy consumption of that request actually comes in.

The initial use is your phone, which is burning power whenever you use it all the time.

One day, maybe AI models will be run on your phone.

But right now, those questions are offloaded

...