Florenta, congratulations on your book coming out TODAY.  Jointly written with Sandra Barbosu, the book is called “Catalysts of Discovery:  How Technologies are Shaping the Future of Science and Innovation.”    Can you tell us about the book?  What are some of its main ideas?

This short book developed from an academic paper with the help of professional editors.  It actually originated in ideas developed in Florenta’s dissertation some years back on the relationships between scientific expertise and technologies for conducting science.   In general, scholars had promoted a finding that, overall, technologies tended to increase scientist productivity, but Florenta and her co-author felt that the story had to be more complicated.  What they found was that technological tools created two types of effects.  First, as anticipated, it increased productivity.  But perhaps more importantly, it also increased the types of ideas that scientists produced.   The direction of innovation was influenced by the tools.  Sometimes scientists worked toward combining knowledge in new ways; sometimes they dug deeper into niches; sometimes they joined the herd and entered new intellectual territory.  It is the capacity of new tools to unleash creativity that really excites Florenta.

As a renowned expert on the impact of AI, you have examined how scientists are using AI to do their jobs, which are of course to come up with important innovations.  What’s fascinating to me particularly is how the nature of the innovations themselves is influenced by the ways that experts use AI.  Can you tell us what you’re finding in your work?

Florenta has done a number of studies in this area over time, which have evolved contemporaneously with a broadening in the definition of AI.   For example, an early work on motion sensing and computer visioning shows how new tools reduce the costs that scientists incur to coordinate with each other.   Computer vision can make it easier to explore broadly. 

In the social sciences, tools such as topic model enable researchers to coordinate across fields.

And in medicine, the emergence of AI may be democratizing drug development by enabling scientists to learn more from relatively low amounts of data on neglected diseases.

Overall, AI is a substitute for the drearier sides of a scientist’s job.   It frees up scientists to do other things.  For example, in our field, STATA – a statistical program that many of us use – makes it much easier to do calculations, which frees us up to do more of the kinds of creative thinking that makes our work distinctive.

Florenta believes that there’s a long runway before these kinds of opportunities become exhausted.  It will take a long time before AI replaces human expertise altogether to the point where scientists become less relevant.

You’ve got a really interesting new paper with Avi Goldfarb on patterns of AI uptake in hospitals.  After the challenges of AI deployment in radiology, what are you finding?  Can we at least use AI to reduce the crushing paperwork burden in healthcare?

There’s a lot of work that’s been done on this question.  Here’s a summary of what Florenta has found:   The capabilities of the technology are not sufficient to translate fully into value for patients yet.   Human concerns have to be addressed given the stakes in healthcare before AI can be fully deployed.   AI did initially fail in radiology, for example.   While it’s much advanced at this stage, there is also a human fear of displacement and reduced quality and insufficient regulation.  These concerns have to be addressed before AI can be deployed fully.   We need deeper understanding of the technology and the incentives that surround it. 

What’s most prevalent now in healthcare is the reduction of the costs of administrative processes, such as the writing of reports and the implementation of various policies.   A recent study in China, for example, demonstrates that triage using a symptom-checking app can both reassure patients and reduce the costs of diagnostics.  There are lots of opportunities for improving workflow through AI.

What are you most excited about regarding the future of innovation?    How is the relationship between human expertise and tool emergence – including both AI and quantum computing – changing the job of the scientist and the output of science?

The tools are complementary to the expertise of scientists.  The direct of research is broadening as expertise is freed from tasks that AI can perform.  In our field, for example, this makes data collection must more efficient.   AI also can be used to run experiments using simulations.  There are lots of applications that have yet to be developed.

What makes you optimistic?

Florenta would love to see an explosion of creativity emerging from the deployment of new tools such as AI and quantum.  Maybe we can even do great good in the world by channeling that creativity toward improving people’s lives.