What's Wrong with AI?

AI’s everywhere, and commenters are asking who pays the real price

TLDR: The blog argues today’s AI boom can drain more electricity and water than most people realize, while companies push it into everyday life fast. In the comments, readers were split between "this is a grounded warning" and "hold on, explain the water claim," with everyone agreeing the stakes are getting very real.

A blog post titled "What's Wrong with AI?" came in hot with a simple message: the writer is not anti-AI, but they are deeply suspicious of the way companies are rolling it out like a shiny toy with someone else’s utility bill attached. The big accusation? Asking an AI chatbot for help can use far more energy than a normal web search, and giant data centers may also burn through huge amounts of water. Translation for non-tech readers: your funny robot helper might come with a side of higher power bills, more water stress, and a lot of corporate shrugging.

But the real fireworks were in the comments. One camp was basically yelling, "Useful doesn’t mean limitless!" with TimByte warning that workplaces and public services are being turned into live experiments. Another crowd praised the post for being unusually calm and grounded instead of full doomsday rage. Student commenter mmilunic brought the existential dread, admitting they feel conflicted using AI tools that can do "literally all the work" while also wondering what that means for jobs and basic ethics. Then came the classic comment-section plot twist: skepticism. skiing_crawling jumped in with a very fair "wait, how does the water thing even work?" question, giving the thread a mini fact-check energy. And burlesona dropped the most dramatic line of all, saying this isn’t a game of chicken — it’s an arms race. Casual!

Key Points

  • The article defines its subject narrowly as current AI tools such as LLMs and image generators, not AI as a general concept or AGI.
  • It states that a ChatGPT query uses roughly 10 times the energy of a traditional search engine query.
  • The article argues that AI data centres significantly raise electricity demand and that most of the energy serving that demand comes from fossil fuels.
  • It says renewable-powered data centres may still indirectly increase fossil-fuel use unless they add new renewable generation capacity rather than relying on existing supply.
  • The article states that AI data centres consume large amounts of water for cooling, with projected evaporation of 4 to 7 billion cubic metres annually by 2027, and says many are located in water-stressed areas.

Hottest takes

"usefulness doesn’t automatically justify unlimited deployment" — TimByte
"it’s not a game of chicken, it’s an arms race" — burlesona
"they do literally all the work" — mmilunic
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.