Friday, January 05, 2024

Comic Cuts — 5 January 2024


Well that was fun. Now, back to work.

I'm still waiting on the latest proof of Beyond the Void so no news there, but I have been busy with the next book. A sort of palate cleanser before I get to the one after next book, I have been working on a new Forgotten Authors volume, updating and rewriting (sometimes very thoroughly) some old essays that I think qualify.

Not that all the authors are completely forgotten. I have been working on two pieces this week, one about early 18th-century writers who wrote about pirates and highwaymen and another about a writer famous for her books about ponies. It makes for quite a contrast. Works by all the authors covered in these pieces are still available, some as eBooks, some through Gutenberg and other free text websites as they are long out of copyright.

And there are collectors of books that will know of the crime and historical writers I'm including, and one is a mystic still discussed over forty years after his death. Are they really forgotten authors? I'd argue that they are to the the vast majority of book readers. I had this same problem with a book that partly shares its name with my books, Christopher Fowler's The Book of Forgotten Authors--but his is a very different book to mine, with short, digestible biographies that skim the subject... mine are long and over-detailed, which is why I fit may 12 or 13 per volume to his 99. When I read the list of authors he covered I kept thinking, "Is (s)he forgotten?" But it's only because I collect old paperbacks that I come across names like Pierre Boulle and Dennis Wheatley with any regularity. To the wider public they really are forgotten.

Most of the works I'm covering are unlikely to be available on Gutenberg or the Internet Archive. They're often mid-list or paperback original authors that will only be found on the shelves of collectors.

They are not the sort of books that would be available to ChatGPT. I have been thinking about AI and the programmes that have been making the news over the past few months. I have had a little play with them, like many others, and the results have been mixed. I am not impressed with ChatGPT's ability to put together an obituary, for instance. The few I asked it to write were basic and often contained wrong information when I came to check the results.

The odd thing is that I suspect that Google's new Gemini AI launch, the basic version of which is described as "text in, text out", will have scraped this very blog. Blogger is owned by Google and, although the content of every post I write is my copyright, I doubt if that has stopped Google using the thousands of blogs it hosts to help its AI learn how to write or where to glean information.

Normally this wouldn't be a problem. After all, I put information out there to share. A link back or a citation would be nice, but I'm choosing to release that information for anyone to use. It's the wholesale lifting of my work that irks me, and that is what Google is capable of.

Similarly, I had a little play around with Midjourney when it came out and generated a few pictures. I didn't go beyond the free allowance and haven't used it since. However, the recent release of a spreadsheet that named 16,000 artists whose work had been scraped in order to teach Midjourney how to draw in certain styles, shows how pernicious AI can be. To quote ARTnews, "During the New Year's weekend, artists linked to a Google Sheet on the social media platform X (formerly known as Twitter) and Bluesky, alleging that it showed how Midjourney developed a database of time periods, styles, genres, movements, mediums, techniques, and thousands of artists to train its AI text-to-image generator."

Access to the link to Google has been restricted, but you can still find the list at the Internet Archive. The list of artists is the third tab and it doesn't take long to do a search and discover that many of my favourite artists are on the list: Chris Foss, Peter Elson, Angus McKie, Chris Achilleos, Bruce Pennington and others who created the covers I loved when I was buying huge numbers of SF books in the Seventies and Eighties. And hundreds of comic artists, starting at the top with Brian Bolland and Dave Gibbons, whose work informs Midjourney.

John Freeman put together a list of over 50 names of people he knew or had worked with. David Roach posted on Facebook that he had been told about the list by his daughter: "Who knows how much of an impact my own work has made on it all. I'm not sure how to feel... possibly a bit annoyed!"

I'd be very annoyed too. You can find plenty of David's work online and more if you dig around Facebook and elsewhere. The Midjourney list doesn't identify what work or how many works have been used for each artist, but it could be in the hundreds of images for some and as few as one for 6-year-old Hyan Tran (see the ARTnews link above).

It has already been decided that an image generated using Midjourney's software could not be copyrighted because of the way it was generated. But that currently means that once the image is released, anyone can reuse it, including the images in this post, which doesn't protect the original artists.

Like many, I'll be keeping an eye on cases as they go through the courts. Here's an article that seems relatively up to date (December 2023), which mentions problems that have faced a class action brought in San Francisco and that a newly amended complaint [linked material downloads as a PDF] was filed on 29 November.

No comments:

Post a Comment