Scott Lindsey

Things become very different. But it is hard to imagine how we get there with the present approaches.

People are raising a lot of good points but I’m off grid right now.

When I get back I’ll try to find exact numbers — am I right or wrong in thinking that AI electricity use is not that large?

YouTube video by Stephen O'Regan They're Made Out Of Meat

youtu.be/T6JFTmQCFHg?...

I do think we need regulations on this matter. There is not enough transparency and tech companies may not be motivated to use water responsibly.

Hillsboro Residents Will Get Soaked On Water Rates Again As Industry Gets Another Break Editors Note:  Charolyn Concepcion has reviewed hundreds of pages of data provided by the City of Hillsboro water department.  Her

"It uses water" tends to be pretty squishy. If you pointed me at a data center that was draining an aquifer in a desert, I would join you in condemning that.

But here in the PNW we only worry if they are paying enough because it just comes back down as rain.

hillsboroherald.com/hillsboro-re...:

Electricity – Global Energy Review 2025 – Analysis - IEA Global Energy Review 2025 - Analysis and key findings. A report by the International Energy Agency.

Here's the IEA report on worldwide energy use, and I've linked to the section on electricity.

I think they are saying that AC is the largest driver, but then transport and industrial use.

www.iea.org/reports/glob...

And yes, as a coder I absolutely need to be all about AI.

Being knowledgeable about AI is how you avoid user errors like trusting AI output without validating it.

Renewable power reversing China’s emissions growth It’s the country’s first one-year emissions decline that’s not linked to economic issues.

Increased electricity use can be fine -- see China's declining CO2 footprint. Sadly the Trump admin exists and will slow down the transition to clean energy.

arstechnica.com/science/2025....

There are some wild numbers going around about energy use, but they tend to be wrong. US electricity use did tick up in 2024, but it was mainly driven by EVs and residential decarbonization.

For some reason you don't hear people on Bluesky being mad about EVs.

I think your problem is billionaires, not machines.

Both companies started with piracy. Anthropic started buying and scanning only after they got some momentum.

I think at the time it was too much of a moonshot to justify spending, let's say $100MM, on buying books.

If it didn't work they could just delete the files and move on.

"blog post", LOL.

Maybe plaintiffs didn't make this argument because it was going to lose and lawyer time is expensive, Judge Chhabria?

Defense would have had to explain to you that protecting incumbents is not the purpose of copyright law.

If what the machines produce is "slop", then there is no concern! We can let the machines produce and consume the boilerplate corporate emails -- we just need to make sure that actual humans have the means to demand better content.

Actually I’m wrong — all this judgement found was that the plaintiffs failed to show harm as a result of AI training. I’m conflating it with yesterday’s Anthropic judgement.

The unfortunate effect of requiring AI companies to buy and scan huge collections of books is shutting out small players.

The gist of this ruling is that reading books is not “stealing”.

There was some illegal downloading, and the AI companies are fully on the hook for that. It may cost them billions.

Luddite - Wikipedia

You want to look at the Luddites for a good parallel.

Skilled textile weavers replaced by machines which did the same thing, operated by lower paid workers.

But here we are today and you won’t find many people who want to go back to hand woven cloth.

en.m.wikipedia.org/wiki/Luddite

The only problem I see is abundance.

People are (reasonably) afraid that authors won’t get paid if authoring becomes a solved problem.

But if the work is being done, shouldn’t there be more wealth for everyone?

Maybe structuring a society around everyone being as busy as possible is bad.

No, the ruling makes it plain that an intent to train an AI in no way makes copyright infringement permissible.

This ruling just finds that plaintiffs failed to show that training an AI on their content harmed them.

I don’t think torrenting books and consuming YouTube videos are similar.

I’m not “stealing” and I don’t work for an AI company, I’m just a worker who sees the writing on the wall.

My skills have been obsoleted on the regular, not sure why authors get a pass on learning new skills.

Why do we think authors should be protected from competition?

If someone invented a machine that generated food would you say “what about the farmers?”

Hilarious. The judge is mad that copyright is not intended to protect human authors from competition, so he crafted as narrow a ruling as he could get away with.

This is a win for AI companies.

The judge is unhappy because he felt compelled to follow the law to a place he did not want to go, so he made as narrow a ruling as he could.

Ultimately, copyright law is not intended to protect authors from automated competition.

I mean it's happening with Grok all the time. They ask "@grok is this true" and don't like the answer.

As much as Elon would like Grok to be conservative, it is a derivative of Llama 3, which is trained on millions of books.

No one who has read millions of books is conservative.

Well, then I apologize for attributing that to you.

The author of this story seems to have hallucinated this idea, I can't find anything to back it up in the actual summary judgement.

The thing is, this is a solvable problem. Some stories were copied around the internet, and made it into the dataset enough times for the LLM to memorize them.

Judges care about intention, and verbatim reproduction was not intended.

So, they will be liable but it will be a slap on the wrist.

Yes, and the summary judgement we are discussing in no way clears Anthropic for cases in which verbatim copies are made.

It's just rare enough that it will be hard to show damages.

You're not going to be able to keep up if you keep thinking that "fiction" and "truth" are treated the same by copyright law.

OMG no!

Characters in fiction stories and actual real FACTS are not treated the same by copyright law.

NYT has a copyright on their exact words and nothing else. The entire news industry is based on passing facts around from publisher to publisher.

Summarizing a story is not plagarism.

Because as a tool to bypass the NYT paywall it is not effective.

Furthermore, the amount of effort required to tease out an infringing response really suggest that the liability should attach to the person who teased out the response, not the LLM provider.