@BlakeHamiltonCA That image is obviously AI generated and anyone should have caught all the problems with it. For water to be that high on Wellington, the Ottawa river would have had to have flooded 80-ish metres high,a disaster that would affect the entire Ottawa Valley. Why didn't you use one of the *many* photos of the *actual* flooding rather than slop one up? Is the article also AI generated?
Replying to @WTL@mastodon.social
@WTL @BlakeHamiltonCA It's especially ironic to use AI-generated images (and content?) to call to highlight climate change issues, given how bad data centres are for the climate!
Replying to @deathbydenim@mstdn.social
@deathbydenim @WTL 100% good point that the image was a bad choice - I rechinyse that now. For the record I didn’t generate it myself, but I did use it, and that’s what matters. I was rushed, tired, and made the clickbait-y choice instead of using a real Ottawa flood photo. I regret that now. My bad.
Though I do think the “AI use = climate hypocrisy” framing is way too black and white binary and oversimplified, though. Data centres can indeed be environmentally damaging, especially when badly located, badly powered or water-intensive. But that is not inevitable. Smarter design, clean electricity/energy source use, better cooling, water recycling/reclamation and avoiding drought-prone regions all matter.
The real issue is not whether or not the technology exists - it does; can’t put the genie back in the bottle. Opposing its existence is like opposing trains for displacing horses & buggies. The real issue with AI is whether we govern and build it responsibly. And we can.
Replying to @BlakeHamiltonCA@mstdn.ca
@BlakeHamiltonCA @deathbydenim On this point, I'll disagree - to a point.
Data centres are necessary. We need legislation requiring them to be built sustainably; green power, and the other points you mention.
LLM/AIs are being sold as end-all for solving every problem, which is completely untrue. They largely answer "what would an answer to this question look like" on a word by word basis, which is why they are so frequently wrong. This is a two or three beer discussion as my dad would say.🖖🏻
Replying to @WTL@mastodon.social
@WTL @deathbydenim That’s a very fair distinction, and I mostly agree. Data centres are indeed necessary; they have been for decades, long before the rise of AI. The question is whether we properly regulate them like serious infrastructure, instead of letting every company cosplay as a public utility with a press release and a hose.
I’m not arguing LLMs are magic oracles. They’re not. They hallucinate, flatten nuance, and can produce very confident nonsense if/when used lazily. But “often wrong” is not the same as “completely useless.” Calculators didn’t remove the need to understand math; AI doesn’t remove the need for human judgment, sourcing, editing, domain knowledge, or basic BS detection.
So yes, it’s not an end-all solution. Not a toy to trust blindly. But also not an evil demon machine. It’s a tool, same as many others…powerful, messy, overhyped, useful, and badly in need of adult supervision.