The UK has a more limited set of fair dealing uses that overlap with the four factors. For example, both cover parody and literary criticism. But the UKs is more of a set of exceptions than a balancing algorithm. I *think* UK law might be worse for AI companies, if they can be sued there. IANAL.

UK law is very like US law. It protects expression. It's more than the literal words—for example, they protect translation rights. But it doesn't cover ideas, facts, style, etc. To judge hard cases, the US has "four factors"—purpose, nature, amount, effect on market—judges are to balance.

If you can get that computer to parrot (regurgitate) a passage of a book, well, you've nailed them to the wall. Even better if that affected the market for your writing. And companies have already been gotten if they downloaded illegal pirated copies, rather than buying and scanning the books.

To your point about copyright law not serving, I suggest you may overestimate what copyright covers. (See my other thread.) Copyright protects expression, not facts, ideas, themes and etc. If it doesn't protect those, it certainly doesn't protect someone using books to teach a computer English.

I'd love to see that compensated in some way. But that seems well within the bounds of what copyright law calls a "transformative use."

There are limits. Claude read a million books and other writings—yours, my wife's, mine and uncounted Redditors—and so it understands English. It was then reinforcement-trained on millions of made-up, copyright-free code challenges. The result: It can write good code when you speak to it in English.

I think regurgitation is going to be a key legal angle, especially for art. The systems cannot help but regurgitate popular characters (see x.com/librarything...). I don't have a crystal ball, but I could see new law coming out of that, ideally also benefitting authors.

At present I don't think you (or my wife) need to worry about AI writing. But artists and photographers are already feeling the pinch. It's not that AI can do good art, but it can do the stuff that a lot of them do for a day job so they can spend their nights doing the good stuff. That's bad.

Ideally I'd like to see copyright return to as it was for most of US history—a limited period, designed to encourage people to write and publish books, not life-plus-75. The Disney/Sonny Bono law discourages creation more than it encourages it. But LLMs are a weird situation and test the system.

I'd like to see copyright adapt. If you want a model, consider how authors get paid per checkout in some European countries apart from and above the price of the book. But I suspect we'll get more gating and more lawsuits and, if the law changes, it will be because Disney is upset, not authors.

We are in weird, new territory, and I hope the law can do better than that. But I am not convinced that training in itself is a violation of copyright and, so far, US courts agree.

I'm just saying you complained I didn't address everything, but not every tweet can! But I linked to where I do at length—boring length for this medium! In brief, I think creatives have reason to be angry, have good cases on several key grounds (e.g., piracy and regurgitation).

I have a lot of concerns about AI. A lot. The impact on education is nothing short of catastrophic so far. The copyright and labor issues are significant. But what's going on is not fake, and its going to have a lot of effects—positive and negative. There are a lot of conversations to be had.

I'm sure you've seen the t-shirts "Librarian: The Original Search Engine" and Gaiman's 2010 "Google can bring you back 100,000 answers. A librarian can bring you back the right one." For years many imagined there was a war between Google and libraries—and good people had to pick the good side.

I'm dating myself, but I've seen several waves. Some objected to @LibraryThing because, even though it uses MARC records and was soon taught in LIS, ordinary people can't "catalog," only librarians. User tags, reviews and etc were *anathema* to many—and are now standard in most library catalogs.

As an author, you must recognize the feeling when someone says they didn't like your book because they wanted X but you wrote about Y.

However, many of your points are why I also posted a link to this lengthy discussion of the topic: bsky.app/profile/libr...

Some of my thoughts on copyright are here. bsky.app/profile/libr...

In essence you reward the model for writing good code—code without syntax errors that arrives at the correct answer. This change from copyrighted data to problems has not really entered popular consciousness, but it's a sea change—and essential to understand.

2. LLMs are based on a lot of copyright text, but AI coding is less so. Copyrighted code is very rare online, and AI coding became a thing because the companies now train models to solve problems.

A few points:

1. "I don't like the change" is not actually a reply to "things are changing." I made no normative statement about coding, still less author's rights. You just think I did.

I'm old enough to remember when people—very much including librarians—would debate the internet. But just as we no longer debate online dating in the same breath as email, AI will become many things and many conversations.

I hope more librarians make this shift, because being an AI denialist just cuts you out of the conversation—a conversation librarians should be in. We already see both major benefits and major problems; to deny either is not to be brave, but ill-informed.

Imagine if, in the space of a few months, librarianship transformed into something else entirely—Broadway choreography or zookeeping, but still somehow accomplishing library tasks, and far better. The change is new and there's much to work out, but if someone downplays this to you, they are a fool.

It's hard to explain to non-devs that the whole career of software development has been transformed. Over the last year AI coding went from "better Google," and "irritating helper" to "HOLY FORKING SHIT!" The real change is less than four months old, since Claude Code. The future… arrived.

The term is barely a year old, but developers often reserve "vibe coding" for when non-devs use AI coding tools, and "AI-assisted development" (etc) for when career software engineers use LLM tools to improve their productivity and scope. Whatever you call it, there is a major difference.

My friend, I'm a career developer who manages a team of four other devs and a devops. This is all the profession has worked on and talked about for a year. So, with respect, I need a popular article from 2025 about as much as you need a Richard Scary illustration of animals playing librarian.

I have seven years more Greek than paleontology, so I'd say PAK-ee-kef-al-o-SAU-rus, or sef, since I say cephalopod with a seph and that's the same word.

I want to understand the psychology of someone making that their floor.

Is "all of us" all society or librarians specifically? How do you think it's being forced on that group? Certainly some employers are pushing it on their workforce—for good or ill, depending—but mostly if you don't like it, you aren't forced.

One could write a haiku on the former:

Cherry blossoms on
Pachycephalosaurus—
Spring is weird this year.

But there's no way to make micropachycephalosaurus fit into a haiku. Or really most poetic lines.