- Index hide
With the amounts of water spent, it would have been possible to build about 370 cars in BMW or up to 320 in Tesla.
AI-focused startups experienced significant funding growth in 2020 and 2021.
OpenAI recently released GPT-4, the next generation of its AI model.
To the surprise of many, a study has revealed that ChatGPT the technology At the moment, it consumes a lot of water, as well as energy. The research has been conducted by the University of Colorado Riverside and the University of Texas Arlington. This study has been commissioned to measure the energy impact of artificial intelligence models such as ChatGPT, GPT-3 and GPT-4.
Microsoft it has consumed water to cool its servers dedicated to AI training, and the impact has been so significant that even the researchers claimed that it has used up to 700,000 liters, and with that amount BWM could produce up to 370 cars, and Tesla up to 320.
The impact of ChatGPT
ChatGPT is a natural language model that is based on AI (artificial intelligence) that is becoming more and more popular all over the world, it was developed by Open AI that has been updated quickly, because it has not been long since this intelligence emerged, which in itself is impressive and has already presented to GPT-4 which is surprising since it can handle more than 25 thousand words of text, which allows to generate, edit and iterate with users in technical and creative writing tasks, such as composing songs, writing scripts or learning a user’s writing style, as well , this new generation surpasses ChatGPT in terms of their advanced reasoning abilities.
ChatGPT consumes a lot of water according to study
The University of Colorado Riverside and the University of Texas Arlington conducted a investigation with the purpose of measuring the environmental impact generated by the technologies to cool their systems and with this the growing carbon footprint left by artificial intelligence (AI) models has been demonstrated, especially the large ones such as GPT-3 and GPT-4.
And it is that the cost of water is also significant when users use ChatGPT since it consumes the equivalent of a 500 ml bottle of water for each chat session of about 20 to 50 questions. An amount that has a great impact if we take into account that millions of people a day use it.
The study shows that training GPT-3 in Microsoft data centers in the US can directly consume 700,000 liters of clean fresh water (enough to produce 370 BMW cars or 320 Tesla electric vehicles) and water consumption it would triple if the training took place in Microsoft’s Asian data centers, but that information has been kept secret. This represents a concern since the scarcity of fresh water has become one of the biggest challenges today.
The researchers also note that companies should “take responsibility and lead by example when addressing their own water footprint.”
It is clear that water is the most valuable thing we currently have, so this impact could have serious consequences if the appropriate solution is not found.
Now read:
Twitter again faces fines from Germany for illegal content
Spotify introduces tool to turn live streams into podcasts
Dalai Lama apologizes after video with boy went viral