Stephen Downes

What bothers me, I think, about some of the schemes responding to the idea that AI is "appropriating" public knowledge and information is that writers are tying themselves in knots trying to avoid the one approach we have always used and which we know actually works: taxation. The basic idea is that if somebody makes money while benefiting off the common wealth - whether it be intellectual property, physical infrastructure, security services, whatever - then they are taxed a reasonable portion of that money that goes back in to support those services. All the other contortions - ranging from 'preference signals' to 'user pay' to the "market-deployment levy framework" - are simply attempts to curtail that democratic process, for example by stipulating, as this article (17 page PDF) does, that our tax dollars should be paid specifically to "creators and rightholders", "public service media organizations", "open access information providers", and "public AI models and services". That's better than most, but why privilege these groups? What's left for the people who sweep the floors where these creators ply their craft?