AI seems to be falling into a pattern of producing one noteworthy story or clusters thereof a week. Which feels about right to keep people abreast of the rapidly shifting landscape without dominating too much.
This week’s AI stories clusters around that tricky intersection between creativity, copyright, and legislation. The slew of companies and organizations willing to negotiate with major AI platforms shows no sign of slowing. Now it’s the news behemoth reuters who have joined the fray. And OpenAI’s Sam Altman has made a key point in relation to this whose importance may be lost on some people and groups that don’t understand how big data works (or just how big the big in big data actually is).
That point is, AI companies desperately need copyrighted material created by human beings to train their platforms on. But they don’t desperately need any one particular set of such material. This came in the context of the ongoing lawsuit with the New York Times. But it applies widely. And its implications matter for writers and the groups who represent us, however much it may stick in the craw to admit why. In short, at the moment AI companies are starting to sit down with and offering money to groups of creative producers. But if those groups play too much hard ball the offer can leave the table pretty quickly and may not return.
The answer that many are seeking is legislation. The Writers Union of Canada is the latest to do so, calling for copyright law to catch up with new technology. But to confuse metaphors in a way no self-respecting AI would, if the genie is out of the bottle then the horse may have bolted before the stable door could be closed. It’s all very well having great protections for future dealings. But in the future AI platforms might not need to deal with too many of us.
AI-Assisted Novel Wins Book Prize
One aspect of copyright law as it stands of course is that it applies to human creation. AI generated content is excluded. The implications of that are increasingly being pushed in practice. This has happened most recently with the news that Rie Kudan has won the prestigious biannual Akutagawa Prize for her AI-themed novel, “Tokyo-to Dojo-to.” The author has admitted not only to using ChatGPT to help, but to pasting around 5% of the book’s sentences direct from its outputs (for a 300 page novel that would be 15 pages of directly AI generated material).
This level of AI content is going to force the discussion of copyright legislation to the fore as countries and legislative regions figure out what kind of protection works like this are afforded.