Earlier this week I finished co-authoring an article with a brilliant coworker about AI and higher education. We both like to experiment in our respective disciplines, and since they overlap, it was fun to work together. Our piece was an attempt to take a sort of snapshot of what's going on at the moment. I imagine it will be outdated sometime this week....
Part of the challenge of fast-moving innovation is trying to rein it in before it does some damage. Last year's onslaught of image generators led to a lot of frustration from artists whose work was being repurposed and infused through various AIs, and now we're seeing the same concerns arise with written work. While the wild-west phase of all this continues-- and continues to define this moment-- I thought I'd check in and see where the conversation around intellectual property is focused. The plagiarism concerns that plague my field (writing instruction) are an outgrowth of this too. Our society highly values creativity in individuals and entities, and as such, we are constantly refining and refiguring its boundaries. Legal discussions usually perplex me. But it's fun and fruitful to ruminate on originality, influence, ownership, and the mysteries behind the creative spark. And that spark is elusive for AI.
"U.S. Copyright Office Rules A.I. Art Can’t Be Copyrighted," Jane Recker for Smithsonian Magazine, March 24, 2022
So this happened about a year ago, following up a ruling from 2019...
"AI-generated comic artwork loses US Copyright protection," Benj Edwards for Ars Technica, February 23, 2023
...and then this happened last week with a work that was generated recently. Legal thought aside, what I find most interesting across these two cases is the timeline. It's a good way to see how the legal premises will continue to emerge. Both emphasize the centrality of human authorship.
"This Copyright Lawsuit Could Shape the Future of Generative AI," Will Knight for Wired, November 21, 2022
Rather than focusing on art, this case has to do with code produced by Copilot, an Ai that generates code. However, the case itself is focused on the mechanism behind Copilot. As it's another form of generative AI, drawing on libraries of code without attributing the source. I find this to be similar to concerns about plagiarism in writing.
"Misinformation, Antitrust and Copyright Loom as Regulatory Hurdles For AI and ChatGPT," Trisha Ostwal for AdWeek, February 21, 2023
All three concerns in the headline transcend boundaries between disciplines, and in addition, across nations. This piece is a good example of concerns that are not just part of US laws, but EU rules as well.
"Will artificial intelligence replace your lawyer–and will its name be Harvey?" Aron Solomon for Fortune, March 1, 2023
Mostly I am putting this here because I look forward to AI lawsuits being settled by AI lawyers. This piece assures us that of course that will not happen, of course this is merely a gimmick, of course this will amplify other issues of access to justice. But if the dataset of these large language models includes a certain Shakespeare quote about lawyers, then eventually one of them might just get an idea...
Comentários