> Firstly, there’s the environmental impact. Models require massive amounts of electricity and water (to cool servers). An AI chatbot contributes more to the climate crisis than a Google search. I really dislike the "people should be better and use less energy" argument for solving macro-problems like this > My naive optimism led me to believe that technology would help us fight climate change. I was wrong: AI and Cr…
> ...why? why would technology that specifically requires a lot of energy help "fight climate change"? There's no good reason it would. Nevertheless, proponents of both AI and crypto have claimed multiple times it would. So I think it is quite fair to bring it up. https://ia.samaltman.com/
I've never been so conflicted about a technology
51–60 of 125 posts
Re: I've never been so conflicted about a technology
#52Earlier quoted context omitted.
[flagged]
(I thought it would be funny - in a self-referential way - to use our especially fine-tuned LLM to analyze bias in the OP in order to prove that LLMs can be useful by showing how they may be usefully applied to the OP itself.)
Re: I've never been so conflicted about a technology
#53I've been using Perplexity, and it annotates its recommendations, as to sources. Maybe it isn't very complete, though.
Re: I've never been so conflicted about a technology
#54Re: I've never been so conflicted about a technology
#55The argument about power usage for developers, as opposed to consumers, is probably insignificant compared to the inefficient use of compute resources for deployed software today.
Arguably LLMs are probably enabling some web developers to create native applications rather than Electron monstrosities saving many P-core CPU cycles multiplied by the number of their users.
Optimising server applications with an LLM could eliminate unnecessary cloud servers.
Of course all the above could be done without LLMs, but LLMs can empower people to do this kind of work when they were not able before.
Re: I've never been so conflicted about a technology
#56> Firstly, there’s the environmental impact. Surely, like everything else in tech, this too shall pass. I expect power requirements to fall away since there is no doubt strong incentives to do so. > My naive optimism led me to believe that technology would help us fight climate change. Yeah, proceeding full steam ahead with planet destruction and praying tech will save us is kind of naive. You're not alone though. >…
AI might make things 1-2% worse, but we weren't suffering from a lack of legitimate content or media. We were suffering from centralized control and walled gardens force feeding us slop in the name of profit, the total enshittification of the internet without regard for the damage done - tragedy of the commons, maybe.
AI could also make things better, because it's no longer worth slogging through the bullshit with search - I'll have AI do my searching for me, or follow its recommendation directly, and never interact with any of the monetized bullshit the search engine tries to pawn off.
SEO content will have to get through ever more competent and discerning AI in order to get eyeballs. That eliminates a broad class of low effort trash, directly demonetizing bad faith actors.
So now you have content creators producing more plausible, but mediocre content, but OAI and other AI providers will have direct access to their logs - if they register someone as having had produced bulk trash content, they can shut down the account, report to authorities or other companies if there's some sort of fraud or illicit behavior going on.
There are competing pressures, and with Google also losing its monopoly on adtech, maybe we'll see companies forced to compete on quality products instead of exploitation of user data, all the while seeing the general quality of the internet improve.
Or maybe AI will just accelerate the race to the bottom and the internet's dead already.
Re: I've never been so conflicted about a technology
#57Earlier quoted context omitted.
> I expect power requirements to fall away This sounds extremely naive to my ears. Performance demands currently outpace resource utilization reductions and it's not clear that we're at the point where that will change soon. Also: https://en.wikipedia.org/wiki/Jevons_paradox > > There are also ethical concerns regarding the methods used to obtain data for training AI models. > This one hasn't registered as problemati…
You mean speeding up the destruction of the internet and media? The problem is it was falling apart before generative AI. This AI has just hastened its demise.
Re: I've never been so conflicted about a technology
#58Re: I've never been so conflicted about a technology
#59> Models require massive amounts of electricity and water (to cool servers). Does it? https://prospect.org/environment/2024-09-27-water-not-the-pr... > training GPT-3 used as much water as just under twice what is required for the average American’s beef consumption. In other words, just two people swearing off beef would more than compensate. https://andymasley.substack.com/p/a-cheat-sheet-for-conversa... > It would…
I agree with your general point, but water usage numbers for farming animals are notorious for being exaggerated. There are studies that take all of the rainfall multiplied by all of the grazing land the cows are allowed on and include that as “water used”. Water usage is also a nuanced concept because water isn’t destroyed when we use it. You have to consider the source and incidentals of processing it, as well as w…
Re: I've never been so conflicted about a technology
#60When I AI code, it's more convenient for me to type in a prompt to change the name of a variable. This involves sending a request to the LLM provider, doing very expensive computations, and then doing a buggy job at it. Even though my IDE can do it for what 0.1% of the energy usage? Or even less? Try running an LLM locally on a CPU and you'll get a glimpse of how much energy it is using to do this simple task.
But coding with AI isn't that huge. What will become huge this year or next is MCP. It will bring "programming" to the masses, all of whom will do stupid queries like the above.
Consider this: I wrote an MCP server to fetch the weather forecast, and have separate tools to get a broad forecast, an hourly forecast, etc. I often want to check things like "I'm thinking of going to X tomorrow. It will be a bummer if it's cloudy. Which hours tomorrow have less than 50% cloud cover?" I could go to a weather web site, but that's more effort (lots of clicks to get to this detail). Way easier if I have a prompt ready.
OK - that doesn't sound too bad. Now let's say I want to do this check daily. What you have to realize is that with MCP the prompt above is as good as a program! It's trivial for an average non-programmer to write that prompt and put it as part of a cron job, and have the LLM email/text you when the weather hits a predefined criteria.
Consider emails. I sign up for deals from a retailer.[1] Now deals from them are a dime a dozen so I've been programmed to ignore those emails. But now with MCP, I can set a simple rule: Any email from that retailer goes to the LLM, and I've written a "program" that loosely describes what I think is a great deal, and let the LLM decide if it should notify me.
Everyone will do this - no programming required! That prompt + cron is the program.
Compared to traditional programming, this produces 100-1000x more CO2 emissions. And because there is no barrier to entry, easily 1000x more people will be doing programming than are doing now. So it's almost a millionfold in CO2 emissions for tasks like these.
[1] OK, I don't do it, but most people do.