1st Conference Report on GAI & Law
On July 29th and 30th, the first cross-disciplinary Workshop on Generative AI and Law (Gen-Law) was held in Honolulu, Hawai’i featuring public sessions with keynotes, panels, and roundtables.
Lots going on with OpenAI’s leadership, and lots going on with the EU AI Act – but on to other things:
On July 29th and 30th, the first cross-disciplinary Workshop on Generative AI and Law (Gen-Law) was held in Honolulu, Hawai’i featuring public sessions with keynotes, panels, and roundtables. GenLaw aimed to unify machine learning (ML) experts and legal professionals to take on the legal challenges posed by advances in generative AI. IP and privacy were the primary focal points, with additional discussions around free speech, liability, and transparency.
The workshop report is worth a full read, but here is a high-level overview::
Report Section 3: Shared Knowledge Base Development
A recurring theme was the necessity for a shared language between disciplines to aid in understanding. The report called for glossaries to bridge gaps between ML and legal terminology, use of metaphors to illustrate complex concepts, and a commitment to stay abreast of the fast-paced generative AI landscape. Refer to the report’s useful Appendix A on cross-discipline terms and Appendix B deconstructions of common metaphors like “the stochastic parrot”.
Report Section 4: Generative AI’s Impact on Law
Generative AI is transformative due to its capacity for broad application, ease of use, and the rapid dissemination and transferability of its products. The workshop underlined the need for lawyers to comprehend the technical nuances, and technologists to grasp the legal significance of their work, advocating for a mutual vocabulary and research agenda.
Report Section 5: Legal Issues Taxonomy
The report identified a taxonomy of legal challenges specific to generative AI, including the difficulty in ascribing intention in legal contexts, privacy concerns due to AI's potential to access and synthesize personal data, and the threat of AI-generated misinformation. Moreover, IP rights are under scrutiny as AI challenges traditional notions of authorship and copyright.
Report Section 6: Research Agenda
The discussions culminated in the identification of research areas where both legal and technical insights are vital. These included the dynamics of centralization versus decentralization in AI development, the evolution of rules and standards for AI practices, the complexity of implementing notice-and-takedown mechanisms (or “machine un-learning”), and the development of reliable evaluation metrics for AI systems.