Editorial - The use of AI

April 17, 2025

If you read anything on this site you can be extremely confident that it was written by a human. There will never be AI generated text anywhere on the site. Unfortunately, however, it requires a large amount of effort and time to stop the content being stolen and used to train AI systems. The scraping systems can be blocked but this needs to be done on an individual basis and it's just too hard to keep up.

AI is used here to clean up photos, using a product called Topaz Photo AI. This system has been trained on billions of photos (part of the user agreement is that Topaz can use images submitted to it for training) so it "knows" about the statistical relationships between colours and pixels. It doesn't change photos in any way except to remove slight imperfections and improve image quality. As an example, low resolution photos taken on a phone have the quality reduced if posted to Facebook. Topaz takes these photos and brings them up to the size and quality required for this site.  Apart from increasing quality, no changes are made to the photos at all.

In an example of a deceptive use of AI, Photoshop has the ability to completely change the background of a photo, but we don't do that sort of thing around here.

Background changed by Photoshop. Guess which is the original.
This is dishonesty.

I paid a lot of attention to AI in the 1970s when the first predictions of computers becoming thinking machines were being made. (One significant book at the time was "Machines Who Think" by Pamela McCorduck.) Very little of it came true, largely because of the state of computer power back then. The ideas were there, put forward by some very smart people (one of the pioneers, psychologist Herb Simon, went on to win a Nobel Prize in Economics; his collaborator, Alan Newell, had to get a doctorate to be employed at a university - he wrote his PhD thesis out of his head in about three weeks and it was passed by the examiners), but the tools weren't up to the job. The interesting thing is that nobody back then thought that AI should be used to replace journalists, novelists, artists or any other creative activity - it would make life easier for them them, but not replace the practitioners. It would take over the mundane tasks like language translation (which it almost has, with emphasis on the "almost) and some areas of research, freeing humans to do the creating. Nobody ever used the term "generative AI".

There are things that computers can do now that were impossible then, but the same predictions are being made about everyone's lives being made much better. As one person observed, however, she would rather have machines do the housework and gardening and allow her time to do the creative stuff like writing and painting, but the pushers of AI seem to want to do the opposite.

The real danger, however, is that as generative AI gets better it will become harder to tell what is real from what is simply made up or just a remaking of things that have been made before. And this is not a good thing.

There are a lot of ways that effective AI can maker life easier for a lot of people, but when all it produces is a condensation and reworking of works stolen from other people it is quite simply plagiarism. (There is a system called LibGen that Meta is using to train its AI models. It consists of the text of an enormous number of books and the authors were unaware that their work was being plagiarised so that a company could make money by regurgitating the content while pretending it is original work.)

---