Wow, that was quite an experience. I recently had the opportunity to testify in a hearing about AI and the future of journalism held by the Senate Judiciary Subcommittee on Privacy, Technology, and the Law. I was honored to be invited to share my thoughts on technology and the news, especially since I knew I would be presenting a different perspective from the other witnesses, who were lobbyists for the newspaper/magazine and broadcast industries and the CEO of a major magazine company. Before the hearing, my fellow panelists kindly acknowledged that they disagreed with my written testimony, which was exactly what I had hoped to achieve. During the hearing, there was limited opportunity for disagreement, as we were only able to speak when spoken to. What struck me the most about the experience was the echo chamber effect in Congress. It’s no surprise that the internet is often referred to as an echo chamber, but the same can be said for Congress, where lobbyists and legislators often agree with each other about the laws they write and promote together. This was evident in a few key areas during the hearing, particularly in the discussion about licensing. Both the industry representatives and the politicians seemed to unanimously believe that AI companies should be required to license and pay for every piece of media content they use. However, I disagree with this notion. I drew a comparison to the early days of radio, when newspapers tried to prevent radio from using their news content. In the end, radio was able to repurpose and share information from newspapers, which ultimately benefited society. I believe AI should have the same right. Some have objected to my analogy, arguing that AI is a program and does not have the ability to read, learn, or have rights like a human. I understand this perspective, but my point was that if we were to argue that Google or Meta has the right to read and learn, it would open up a whole other discussion. The main point is that if AI creators are required to license all content they use, it grants them fewer rights than media outlets and journalists, who constantly use and repurpose information from each other and sources. I do believe there is a distinction between using content to train a model and using it in the output. While it may be fair for large language models to be taught the relationship between words, such as “White” and “House,” I believe it is a fair discussion to separate questions of proper acquisition and terms of use when an application quotes copyrighted material from behind a paywall in its output. The magazine executive cleverly conflated training and output, arguing that any use of content requires licensing and payment. However, I believe this sets a dangerous precedent for news media itself. If all use of all content requires licensing and payment, it could have a negative impact on the doctrine of fair use and the ability of journalists to do their job effectively. Overall, the hearing was a surreal experience, and I hope that my alternative perspective was able to shed some light on the issue at hand.
