I had the pleasure earlier this week to speak at the All In AI Event in Montreal. This was a premiere 2 day event described as “The most important event dedicated to Canadian AI”. The programming was truly sensational featuring the “whose who” of Canadian AI with a wide range of topics including a series on responsible AI, and AI and its impacts on cultural diversity.
The event was led off by several prominent politicians including ISED Minister François-Philippe Champagne who announced the Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems.
My keynote speech was titled “Creating Tomorrow Today: AI, Copyright + the Wisdom of Experience”. In summary, my speech focused on the following:
- The enormous contributions generative artificial intelligence (GenAI) can make to cultural diversity including by its capabilities to generate content, translate content, educate, provide a fair representation of culture and reduce bias.
- Providing examples of GenAI and recent U.S. rulings that GenAI content cannot be protected by copyright and the debate in the creative industries as to whether GenAI content should be protected by copyright.
- The reactions of the cultural sector to GenAI produced content and in particular the criticisms that GenAI systems have been trained without the consent of or compensation to copyright holders, that human authors are being replaced with machine generated content, the unfairness that authors are being forced to compete with AI generated content that is trained using their works, and the potential adverse effect on culture if human created content is replaced by recycled content generated by “smart machines”.
- The 9 class action lawsuits launched by copyright holders alleging infringement in the photograph, software, web, and book markets. I explained the types of claims being made against the GenAI platforms and their defenses including that training AI algorithms is protected by fair use in the U.S. and that GenAI outputs are not infringing derivative works.
- How the history of technology – from the printing press to gramophones to broadcasting technologies to the Internet and online piracy – shows that copyright laws have adapted to balance the uses of innovative technologies with providing protection to copyright holders. These developments have been fostered by progressively adapting treaties and conventions and multilateral agreements to address challenges to authors including the Berne and Rome Conventions, the WIPO Copyright Treaties, CP TPP, and USMCA agreements.
- While history shows that laws can adapt, it also demonstrates how slow changes can be. For example, it took Canada 16 years to implement and ratify the WIPO Copyright Treaties. Moreover, progressive legislation is becoming much more difficult to enact in light of the disinformation power of digital and social media with recent examples being ACTA and PIPA/SOPA in the U.S. This trend is worsening with AI powered disinformation and will soon be even worse with AI systems powered by quantum computing.
- The international initiatives to provide authors with some control over how their works are used in GenAI systems including the proposed transparency rules in the EU AIA, the recent Bill introduced into France’s National Assembly, and the UK’s proposed (and then dropped) plans to introduce a broad text and data mining exception for training AI algorithms.
- I closed by noting that Canada’s proposed artificial intelligence law, AIDA, is silent on the copyright challenges associated with AI.
My slides are shown below and can be accessed at this link.
There were many interesting talks dealing with AI and cultural diversity, copyright and AI, and responsible AI. The speakers included these individuals:
Roanie Levy, Jérôme Payette, Ian Chai, Pauline Halpern, Céline Castets-Renard, Octavio Kulesz, Caroline Jonnaert, Sandra Rodriguez, Céline Mornet, Julia Kastner, Alexandre Teodoresco, Sophie Fallaha, Luc Vinet, Bernard Sinclair-Desgagné, Frantz Saintellemy, Sarah Gagnon-Turcotte, Olivier Blais, Phil Dawson, Duncan Cass-Beggs, Philippe Beaudoin, Valérie Becaert, Mathieu Marcotte.
AI All-In Event Slides final