artificial intelligence | News Story

27 September Webinar: The AI Orphan?

2 September 2024

Protecting science and research in an AI debate dominated by the entertainment industry

See the recording here

When?

Date: 27 September, 2024
Time: 15:00-16:30 CEST / 14:00-15:30 BST

What?

This webinar will outline a number of the pressures that publicly funded institutions and researchers face in the current AI environment and suggest a pragmatic way forward.

Save Your Spot!

Register now to secure your spot!

The Webinar

*Image by Antonio Guillem, licensed from Shutterstock*

Despite the potentially revolutionary application of AI in the scientific, medical, and technological fields, recent discussions on copyright and AI have focussed almost exclusively on the entertainment industries. As a horizontal body of law, copyright however regulates scientists as much as singers, doctors as much as dramatists. We therefore face the prospect of laws being developed for one set of stakeholders, to the detriment of scientific advancement and data-driven innovation.

With the exception of a handful of countries with secondary publishing laws, Europe has a uniquely unfriendly copyright regime when it comes to scientific research. While policymakers are conscious of the innovation paradox – a situation where Europe has relatively high levels of public investment in research but struggles to capitalise on this in terms of commercial rollout – they continue to ignore the harmful impact of the lack of flexibility copyright provides to its scientific researchers (and by extension its research-intensive industries).

This webinar will outline a number of the pressures that publicly funded institutions and researchers face in the current AI environment and suggest a pragmatic way forward.

Particular concerns include the chilling effect of pressure from STM publishers during the negotiation of university annual subscriptions, attempting to limit researchers using what their university has purchased access to for AI purposes. This has resulted in a number of public statements from library consortia and universities around the globe.

Another is the contradiction between the EU’s stated goal of supporting knowledge valorisation and university spin outs, and the barriers to this created by the CDSM Directive. Unlike jurisdictions such as Japan, Singapore, Israel and the US, Art 4 of the Director limits what a model can train on and frustrates university spin outs and knowledge valorisation. As a result, despite the EU AI Act seeking to promote safe AI, limiting what a model can train on promotes bias and poor model prediction. Finally, there is the under-use of the unique collections of public institutions that can improve the quality of AI models. These institutions have a unique position in the scientific and AI ecology, and policymakers need to harness and build actively on their strengths, as well as to make the most of the tens of billions of tax payers euros spent each year on science and research.

Why You Should Attend

In the context of mounting pressure on policy makers by the entertainment industries to restrict AI training and use, this webinar highlights the perspectives of the research sector when it comes to AI regulation. It provides participants with an opportunity to understand more about how to make the European copyright framework more supportive of scientific research in the new mandate of the European Commission and Parliament. Crucially, it is a chance to reverse Europe’s current approach, which stands to promote bias, and poor model prediction. This is something which policy makers need to pay much more attention to.