[eamonnsullivan]

The one thing that had me reading to the end was the mention near the top of the environmental impact of LLMs, but he never got back to it.

I'm currently writing an onboarding doc for my team, encouraging LLM use for some tasks. (OK, well, I'm actually procrastinating by reading HN).

At the same time, I'm in a darkened office with tinfoil on the windows and a fan pointed at me because it's hell outside and it has been for weeks, and every year it seems to get hotter and hotter and we have longer and longer heatwaves.

This seems ... discordant, at a minimum.

Really, _should_ we be using these things to speed up, say, dependency updates if the cost is the planet? I wanted to know what the author thought about that.

[cognitiveinline]

Your commute is worse for the environment than your day's token usage, by a magnitude.

[bluefirebrand]

Hey, I'm happy to work from home. I'm not the one choosing to have a commute, companies are forcing that.

Wait a sec, aren't companies also making LLM usage compulsory?

Wow it's starting to sound like they don't care about the environment!

[eddyzh]

This checks out 50-250kg CO2 are common guesses for 10k spend.

That would be in the same order of magnitude of an average daily commute with an older car.

[eamonnsullivan]

Well, I doubt that. I mostly work from home. But also I live in a big city (London) with a working public transport system.

[cognitiveinline]

I dont know what to tell you. I guess - correct your misconceptions?

Carbon footprint of commute is in KGs of CO2. A day's worth of AI, depending on model, is ~15-100g of CO2.

Heck food you eat has more than a few Kgs of footprint everyday.

[eamonnsullivan]

I won't belabour the point, but I think someone needs to correct their misconceptions. You're comparing a single, average person's usage of AI in a day -- not the billions who actually use it. Then you are comparing it to an American's two-hour commute in an oversized SUV. (A commute on the Tube costs about 0.02991 kg per passenger, per kilometre.)

What about training those models? Or the usage of all of the data centres? The projections are that by 2028 a fifth of all energy consumption in the U.S. alone will be for AI.[1]

[1] https://www.technologyreview.com/2025/05/20/1116327/ai-energ...

[paulhebert]

Does that include the training?

[cognitiveinline]

yes. training is a magnitude lower of energy of the lifecycle of frontier models. with current use of opus etc, the math is closer to 1%.

https://libguides.usc.edu/blogs/USC-AI-Beat/hidden-costs-of-...

[eamonnsullivan]

This is actually a good link. Thank you. It's usage that is the big driver, and LLMs are a big driver in emissions and other damage to the environment.