Accidental Tech Podcast677: I Accept the Battery Cost
https://atp.fm/677
If you really don’t like AI, we have some bad news for you.
Machiel@atpfm I know.. im sick and tired of AI everywhere and all the talk about it.. but I get it’s here to stay
@atpfm it’s just.. we just don’t think about the energy use, all the stuff that was stolen to make it, and how it’s used to mess up society even more. Just because ‘he it works for me’
John Siracusa@vmachiel Many people do think about it, including me! And we all should.
Colman Reilly@Colman So will you, and everyone else, eventually—for better AND for worse. (See also: electricity, automobiles, modern agriculture, manufactured goods of all kinds, etc.)
@siracusa i do not like the comparison with electricity etc i must say.. those provide huge benefits for everyone. AI uses so much power and stolen material to provide benefits for a relatively small group of the population. Electricity made us all beter and more equal, AI is making it worse.
@vmachiel I think your opinion of electricity would have been very similar to your opinion of AI had you lived during its dawn! Electricity initially "provided benefits for a relatively small group of the population" while its generation poisoned the air and polluted the water, all of which affected the masses way more than the elites. Oh, and eventually…climate change.
@siracusa but adding AI on top of our current energy needs is just irresponsible. At this point. Even without AI we are damaging the planet beyond repair. We are burning to planet to generate fake narratives.
Plus we didn’t steal everyone’s stuff to make electricity. And electricity doesn’t lie.
AI is a curiousity, not a necessity for a decent life like electricity is. You want to use it, but the whole ethical justification is flimsy. And AI is not inevitable
@vmachiel Electricity was also initially seen as "a curiosity, not a necessity." It was also not seen by everyone as "inevitable." Hindsight is 20/20.
As for "stealing," we have many precedents, both good and bad, in this area, from the printing press to player pianos to VCRs to fonts to APIs. Sometimes we come up with a pretty OK system, and sometimes we end up making things worse. But the basic questions of ownership and compensation in the face of new tech is very old and recurring.
@vmachiel And as for the environmental impact of AI's energy usage, despite the fact that it is somewhat overblown as a percentage of total human energy use, it surely will be a short-term problem. And I'd argue that crypto mining (a much more clearcut case of harm with little practical use) shows we are not good at dealing with things like this. But advances in *other* tech (renewable energy) does give me some hope.
Secundus@siracusa @vmachiel Energy usage is also often overstated. Nobody is arguing that AI does not consume energy, or that aggregate demand does not add up. But this has to be evaluated in relation to the use and that it is spread across hundreds of millions of users. When broken down per individual and compared with what a typical Western European or American uses energy for, AI usage is negligible.
Niek van de Pas@secundus
I would be interested in references if you have them. I’ve been having a hell of a time finding numbers on this stuff.
I would be interested in references if you have them. I’ve been having a hell of a time finding numbers on this stuff.
@niekvdpas As far as is known, AI companies do not publish detailed data, so there are more or less plausible guesstimates. Andy Masley (see link) has written several solid articles on the topic. Granted, he is just another person on the internet, but I find his argument reasonably convincing. This is not about specific numbers, but about a rough estimate of the order of magnitude of LLM energy consumption and how that compares to overall personal energy use
https://andymasley.substack.com/p/a-cheat-sheet-for-conversations-about
@niekvdpas To give an even vaguer answer. A fairly limited LLM can be run reasonably well on my four-year-old laptop. A much more capable model can be run on a machine, such as a Mac Studio. This does consume energy, but not very much. Especially given that typical LLM use does not imply sustained full-load operation.
Compared to these examples, optimized hardware or software, scale, better utilization, etc. will be more efficient.