Pouring out a gallon of water is an immediately identifiable thing
Yes, as in “you can impress people with that”, no, as in “it gives any sort of accurate perception”
There are plenty of things that consume orders of magnitude more water than AI. You’ll be much better off ecologically by skipping the last few seconds of your shower or eating a tiny slice of meat less every week than by not using these tools.
Besides, many of them can be run on your regular, air-cooled PC.
That’s not to say that AI training in general is not an ecological issue - but most of it actually goes in directions other than LLMs and image generation tools, and for models already trained, the footprint is straight up negligible. You likely wasted more energy reading this than getting an AI to make an image for you.
Yes, as in “you can impress people with that”, no, as in “it gives any sort of accurate perception”
There are plenty of things that consume orders of magnitude more water than AI. You’ll be much better off ecologically by skipping the last few seconds of your shower or eating a tiny slice of meat less every week than by not using these tools.
Besides, many of them can be run on your regular, air-cooled PC.
That’s not to say that AI training in general is not an ecological issue - but most of it actually goes in directions other than LLMs and image generation tools, and for models already trained, the footprint is straight up negligible. You likely wasted more energy reading this than getting an AI to make an image for you.