andymasley

I based the data center numbers on the updated post, not the retracted post

Why then is there a general negative correlation between the cost of potable water and the population of an area in places where freshwater is plentiful? There's a lot of literature on water economics and this pattern usually holds.

Well that's why I cover things at the national, local, and personal level separately. I think if this were any other industry using 2% of a county's water you wouldn't have much of a reaction. In most countries only ~20% of the water is used by households, the rest goes to farms, industry, and commercial buildings

Just as an example, Loudoun County has the most data center concentration in the world by a wide margin, and also has significantly lower water prices than the Virginia average. Seems like they've been able to build out huge new water capacity for data centers and that hasn't raised househould prices, or even kept them low. https://andymasley.substack.com/i/171855599/the-county-with-...

The dictionary definition of useful is “Able to be used for a practical purpose or in several ways.” How am I going against this definition?

This is all so strange. You're reading the exact opposite meanings into my really simple posts and comments. I don't know how anyone could read the post and come away thinking I believe anyone questioning chatbots thinks they're completely useless. I'm gonna bounce. Good luck.

Basically everything I think is here, I think this is all pretty simple and straightforward: https://open.substack.com/pub/andymasley/p/ai-can-be-bad-ove...

If you bought a hammer and never used it, so it never actually improved yourself, would you say the hammer itself isn't "useful"?

I'm really confused about what you're reading into this. I don't "equate" nuclear weapons with video games, I say "here are two completely unrelated things that I'd consider net bad overall, but useful in specific places." Would you say I'm equating guns with balloons if I say they're both man-made? It's hard not to think you're intentionally misreading this.

Yes, it is possible for something to be useful in specific circumstances but still be bad overall.

We have disagreements about what counts as useful. If our definition is "This is only useful if it leads to longterm happiness" that seems way too specific and would exclude too much.

It's stupid to cheat, I agree and try to make that clear. What I'm saying is the claim "Students think they're learning when they cheat using AI" assumes students are so stupid that they think cheating off of a robot will help them learn as much as writing an essay themselves. That's obviously wrong.

I think basically everyone using tobacco knows it's bad for them. They're not stupid. This is another example of people being basically aware of their situations.

I'm not making a claim about whether they're good overall or not, I'm saying 1) Reasonable people can think they're bad overall. 2) It's not reasonable to say they're literally always useless.

Sorry this is pretty straightforward. If I'm Kim Jong Un, a nuclear weapon is extremely useful to me. That doesn't mean I think overall a world where nuclear weapons exist is good. I'm confused why you think "useful" needs to have this additional meaning of "good overall"

I mean I say in the post I think it's bad that people are using LLMs to cheat.

I think each are net bad and shouldn't exist yes, but I also think each is useful in specific contexts. Not sure why that's not relevant, it's a direct example of stuff that's useful but net bad.

Sorry the energy comment is ridiculously out of context. I've written a deep dive on how small that number is. Do you complain when YouTube videos or video games use similar amounts of energy? Your laptop uses the same energy every 3 minutes. https://andymasley.substack.com/p/a-cheat-sheet-for-conversa...

There's a ton of stuff I think is useful in specific circumstances but can be bad overall.

-Video games: Provide fun, but probably overall bad for society bc people waste too much time on them.

-Alcohol: Most drinkers get a lot of value out of drinking, but alcoholism is so bad that on net alcohol's probably bad.

-Guns & nuclear weapons: Wish both didn't exist, but each provides a lot of use to the specific people who have them.

-TikTok: Overall causes too many people to believe misinformation, but for a lot of other people is fun or interesting.

It's possible to think AI chatbots are net bad because people use them to cheat, or they rely on them for information too much and believe false information, without believing that they are always useless in all circumstances. I can use ChatGPT to alphabetize a long list for me. That's useful, even if I think overall chatbots are net bad.

I say in the post I lose respect for people who specifically claim that a billion people are using an app that adds absolutely nothing to their lives each week, not people who dislike chatbots for other reasons (hallucinations etc.). So I think a lot of people are getting a lot of misinformation from TikTok, and I think it’d be better if TikTok didn’t exist, but I’d consider anyone who said that TikTok is completely useless to its users to be pretty goofy. I feel the same about chatbots.

I say in the post I lose respect for people who specifically claim that a billion people are using an app that adds absolutely nothing to their lives each week, not people who dislike chatbots for other reasons (hallucinations etc.). So I think a lot of people are getting a lot of misinformation from TikTok, and I think it’d be better if TikTok didn’t exist, but I’d consider anyone who said that TikTok is completely useless to its users to be pretty goofy. I feel the same about chatbots.

Hi I’m the author, pretty confused about how you can come away with this reading. I say at the end that a lot of AI criticism is correct, there’s just this specific type I think is lazy

Hi I’m the author, I agree people believing AI hallucinations is a big problem and will edit the post to clarify

I added in the article that I think this is all the wrong way to think about environmentalism anyway and that people should spend most of their time working on systematically changing the energy grid to renewables. But I keep meeting people who are worried about their personal emissions from ChatGPT, so I wanted to make it clear that if you are worried about that ChatGPT isn't really a drop in the bucket compared to other things you could change.

The point of the article WAS to debate someone criticizing something they don't understand. AI as a whole is using a lot of energy and we should think about the environmental impacts, but I pretty regularly meet people who think that every individual ChatGPT search is uniquely bad for the environment. I tried to make it as clear as possible that that's the issue I'm responding to, not all AI energy use.

I wrote the article. Water use is used really ambiguously by people talking about data centers and LLMs in general. To get the numbers I just took the estimated water used per kWh (https://www.sunbirddcim.com/glossary/water-usage-effectivene....) and multiplied it by the number of kWh used in data centers for different tasks

If we're trying to figure out a reasonable number for how much energy a single ChatGPT search uses, it seems weird to factor in all future training of all future models. It would be like trying to figure out the carbon cost of a single plane ride and trying to figure out what fraction of additional plane flights that one ticket incentivizes. It's too murky to put a clear number on and probably doesn't add much to the cost on its own. I tried to make it clear that the post is about your own personal use of ChatGPT, not the entire AI industry as a whole.

I specifically mentioned in the article that I think it's mostly a waste of time to worry about your personal carbon footprint and you should aim for big systematic change instead, but because so many people seem to be worried about their personal chatgpt use I wanted to clarify that it doesn't use much energy anyway

I am to put it mildly not an e/acc and referenced being very worried about other risks from advanced AI in the article.

Most of that money went into infrastructure + human talent, with energy costs at the tail end. I'd be happy to include the energy costs of producing the materials, feeding the humans and moving them around etc. but those aren't usually included in environmentalist critiques of the energy used in training. I couldn't use the original article to unpack every possible interpretation of how much energy was used without it turning into a book. If we did the same for plane rides we'd also need to include the energy cost of assembling the planes, training and feeding the human pilots etc.

If I'm going to be accused of rationalizing my laziness I need at least one source saying my numbers are off. I'm going off the numbers provided by environmentalist critics of ChatGPT. I hope that by sharing my post people feel exactly as guilty about using ChatGPT as they do about buying one additional digital clock.