Climate change is firmly back on the news agenda as more and more temperature records tumble worldwide.
Once again, scientists are reiterating that there is robust evidence of the link between CO2 pollution resulting from our consumption of fossil fuels and our warming world.
At the same time, generative AI has pitched up. ChatGPT has become one of the fastest-adopted technologies ever. But ChatGPT uses a phenomenal amount of energy. The Indian start-up Zodhya has calculated that training ChatGPT-3 would have required over 1,000 megawatt-hours (MWh) of electricity, which is enough by some estimates to power one million homes annually.
And that’s just to get started. Each ChatGPT query is said to require 6.79 watt-hours, the equivalent energy of running a five-watt LED bulb for one hour and 20 minutes. And we know globally there are a lot of queries being made and, therefore, a lot of energy is being consumed.
