Today, I spoke at the the U.S. Senate's AI Insight Forum, which focused on transparency, explainability, intellectual property, and copyright. Click here to download a pdf of the testimony.
Majority Leader Schumer; Sens. Rounds, Heinrich, and Young; and fellow attendees:
I am grateful for the opportunity to participate in this important forum. I believe it is key for the legislative branch to be engaged early in considering the governance challenges of transformative emerging technologies like artificial intelligence. Indeed, much of my work has been focused on strengthening Congress’s capacity on emerging technologies, including building in-house technical expertise, engaging in horizon scanning, and assessing the full spectrum of economic, strategic, and societal implications of new innovations.
My organization, the Foundation for American Innovation, has a mission to promote “technology, talent, and ideas that support a better, freer, and more abundant future.” Our work is rooted in an optimistic vision for the future, in which technology is aligned to serve human ends: promoting individual freedom, supporting strong institutions, advancing national security, and unleashing economic prosperity. We believe that better collaboration between innovators and policymakers is vital for making this future happen sooner.
Our worldview holds that technological progress has been an overwhelming net positive for humanity. Yet, we also recognize that new technologies sometimes bring novel harms and disruptions, undermining our security, our livelihoods, and our social and community life. Rather than only focusing on the upsides of innovation, we believe dynamism is best advanced by working in good faith to anticipate and mitigate these challenges.
While my colleagues have done significant work on issues like AI attribution and explainability, the topic I’d like to focus on is the intersection of AI and copyright. In the case of AI, there are important policy questions at both the developer and consumer ends of the technology. These include the applicability of fair use for model training data, the copyrightability of model outputs, copyright-adjacent issues such as right of publicity, as well as impact on the content industry and its workforce.
While these may seem daunting, AI is standing in a long line of digital technologies that have threatened to disrupt content-centered industries. These include video and music streaming services, digitized books and libraries, and search engines. Each, more or less, has been resolved in a way that balances the interests of rights holders and consumers, bringing significant benefits.
AI is on a trajectory to enable an unprecedented explosion of both artistic creativity and scientific progress. While there are some challenging edge cases, many issues can be handled by existing law (and multiple cases are currently moving through the courts). As Congress considers these issues, any new legislation should be rooted in a constitutional understanding of our system of intellectual property as an incentive for creativity and innovation. That is, “to Promote the Progress of Science and useful Arts.” This big picture view should be prioritized over the narrow interests of any particular rightsholder group.
More broadly, artificial intelligence is a dynamic and rapidly evolving field. Each month brings powerful new tools, new startups, and critical research. At the time of this meeting, OpenAI’s ChatGPT will not even be a year old. At this early stage, the complete range of opportunities and use cases is not well grasped. Even if well intentioned, overly risk-averse regulatory intervention could block many beneficial applications and blunt America’s strategic competitive edge.
Policymakers should prioritize narrow interventions in well understood areas, as well as building deeper expertise and oversight capacity at agencies and in Congress. For the latter, this should include investing in additional committee staff, increasing funding for support agencies like the Congressional Research Service and Government Accountability Office (and its work to assume the mission of the Office of Technology Assessment), and expanding its relationship with NASEM (including streamlining the contracting process). They should also find ways to strengthen AI expertise in the Copyright Office and the Library of Congress generally.
Thank you again for the opportunity to participate in this forum. Speaking for myself and the FAI team, I look forward to future engagement with you on this topic.