On November 12 and 13, I had the opportunity to participate in the Pistoia Alliance’s 2024 conference in Philadelphia, PA. The Pistoia Alliance is a global, not-for-profit members’ organization focused on pre-competitive collaboration to tackle problems faced by many in the life sciences R&D space. CCC has been an active member in the Pistoia Alliance for many years.
Unsurprisingly, Artificial Intelligence (AI) and data management were topics in heavy rotation at this year’s conference. I was pleased to see that the attention paid to AI had moved past the general hype we have all been observing over the past few years, and instead was focused more on the practicalities, realities, and limitations of what AI has to offer.
Why AI?
Why is AI being discussed in the first place (beyond the general interest in AI from all sectors)? Several presenters at the conference set the broader context by citing well publicized statistics on industry investment and industry success. Essentially, the growth in the money spent on drug discovery has far outpaced the growth in new drugs discovered, and the pharma industry is still struggling with how to narrow this investment-new drugs gap. AI is seen as a solution for closing this gap: increase the number of new drugs or uses discovered while reducing the duration of the cycle and cutting overall costs.
Despite the promise and emphasis on AI, there were repeated efforts to play down the hype and expectations and to adopt a more sober and comprehensive view of AI’s role and benefit for drug discovery—there is a pressure to adopt AI that risks distracting people from the core problem that they are trying to solve. There was also a recognition that AI is more than the application of a specific technology. Organizations need more to be successful with AI. From the presenters’ point of view, this means primarily investment in people, but also good data management, data governance, and data and computing infrastructure.
Fully AI Ready
A big part of good data management and data governance for AI is having a plan for FAIR (Findable, Accessible, Interoperable, Reusable) data. FAIR was a prominent topic at the conference in part because Pistoia has many long-running, related projects underway for its members, including the FAIR Community of Experts focus on implementation of the principles. Ultimately, however, the industry still struggles with what to do with FAIR.
One question asked at the conference was: why is it so hard to sell the value of FAIR after all these years? Importantly, this question did not represent a loss of faith in the value of FAIR, especially for AI. Quite the opposite — most of the Pistoia attendees believe that FAIR is a necessary and beneficial pillar for effective AI. Rather, the problem is seen as one of education and communication; how do we quantify and communicate the value of FAIR to the organizations needing to adopt it and invest in FAIR data? Here, too, Pistoia Alliance is pushing the conversation forward, both at this conference and with their various FAIR implementation projects.
Copyright and AI
Another important consideration for AI readiness — in addition to investment in the right people, infrastructure, data management, and governance — is, of course, copyright. During this conference, CCC sponsored LinkedIn polls to survey conference attendees on their view of copyright and AI. What did we learn?
- 82% of respondents either “sometimes” or “often” use generative AI in their research
- 67% of the attendees take copyright into consideration when thinking about AI
- Most organizations do not have a dedicated team or person to address copyright and AI
- Most respondents are concerned about different AI regulations with cross-border collaboration
As with communicating the value of FAIR principles, we see a similar need to continue education and communication around the role of copyright and AI. One place to start is learning about the intricate intersection of copyright law and large language models (LLMs) and what that might mean for your own applications of AI.