Industry engagement offices begin early adoption of generative AI tools for efficiency, marketing
Excerpted from the February 2024 issue of University-Industry Engagement Advisor. UIDP members can view the entire issue here.
Generative AI may be reaping large headlines and gaining wide acceptance and application in numerous industries, but when it comes to the professionals who lead university corporate engagement efforts, they are still very much in the learning stages. In fact, more than a dozen industry engagement leaders contacted to comment for this article demurred because it was just “too early” at their university.
Still, the interest is clearly there, and fortunately there are several experienced individuals — especially within university tech transfer offices — to whom they are turning for advice as they start their journeys. What they are learning — both from these experts and through their own experience — is that there are a growing number of tools and applications from which to choose, even within a single provider’s offerings. Second, there is no substitute for just getting in there and testing these tools. It’s the only way, they’ve learned, to get better at giving “prompts,” to recognize “hallucinations,” and sort out the helpful use cases versus the time wasters.
As for how AI can contribute to the strength of a university’s industry engagement efforts, the possibilities are virtually limitless — from simple, time saving uses such as writing “form” letters, to improved and more efficient contracting, to better identifying and targeting potential industry partners.
For example, Emily Hostage, director of industry relations with Columbia University, is leading a task force, working with individuals from licensing, contracts, compliance, finance, and others on how best to apply AI to improve her office functions.
“With business development, first-pass human generative may become AI-generated,” she observes. “In doing marketing summaries, getting new invention reports, you need to understand the industries to which they apply and to what companies,” and AI tools could become a faster and more comprehensive way of figuring out those connections.
Hostage adds a caveat (one of many she and others share) that for now, most versions of ChatGPT are “frozen in time” — and that “time” is not today, making results potentially incomplete as well as out of date. “It may be 12 or 18 months ago,” she notes. “I suspect that will change in the future, but a lot of products do not change that rapidly.”
With current tools, she continues, one potential use is putting a prompt into ChatGPT like ‘Please read this disclosure and tell me what markets it applies to.’ “It can present a pretty decent first draft, automatically collecting a roster of companies selling cancer products, for example,” she offers. “We’ve done [prompts like] ‘Give me the market share for the top five companies selling drugs.’”
Overall, she says, “We’re running about 5-10 [AI-assisted] projects, testing what works and what doesn’t,” then preparing internal briefings and sharing results.
On the other side of the country at the University of California San Diego, “we’ve been learning more about GPT models in the past year or so — particularly from the research perspective, because we have faculty working on these types of things,” says Cody Noghera, chief corporate relations officer in UCSD’s Jacobs School of Engineering. “As far as making use of the tools, there was a pilot launched just in January, where the campus was asking people with different jobs the areas where they use it, and how it responds to prompts — around specific UCSD knowledge bases.”
“We don’t share the ‘secret sauce’ in how we find what we find, but we certainly use different AI tools to figure out who to target,” adds Marc Sedam, vice president of technology opportunities & ventures at New York University and a recognized “guru” of AI in the university commercialization world. “In some ways it’s a glorified search engine, but certainly these tools are able to sift information better and require a lot less work. One great example is finding things like market size, and who is in [a market]. In the last six months, with AI-enabled tools we’ve been able to summarize market information so much more quickly and cleanly, and that helps us with who we need to target.”