AI is becoming indispensable, yet it is not without consequences for the environment – quite the contrary! The training of models requires very high-performance equipment, which has a significant environmental impact in terms of its manufacture and is very energy-intensive. Furthermore, a great deal of energy is required for our prompts/queries, and moreover water, for the cooling of the data centres used.
For this reason it is important to critically asses the need for and use of generative AI. Its use can be avoided when an alternative is available. At the very least, critically assess the different available generative AI models in terms of their environmental impact during their construction as well as their implementation. It should be noted that an AI trained in a data centre has a different GHG impact according to the data centre’s location.