Posted by
Ideas: Exploring AI frontiers with Rafah Hosn
Energized by disruption, partner group product manager Rafah Hosn is helping to drive scientific advancement in AI for Microsoft. She talks about the mindset needed to work at the frontiers of AI and how the research-to-product pipeline is changing in the GenAI era.
On Oct. 29, the Royal Society announced that Microsoft Research’s Luca Cardelli has received a Royal Society Research Professorship. This prestigious post provides long-term support for internationally recognized scientists of outstanding achievement and promise, and previous holders of the professorships include Nobel Laureates and Presidents of the Royal Society. The announcement also includes the University of Oxford, where Cardelli will be spending part of his time.
He also was in the news recently for his work—along with colleagues from the University of Washington; the University of California, San Francisco; the California Institute of Technology; and Microsoft Research’s Neil Dalchau and Andrew Phillips—in inventing a programming language to build synthetic DNA. After that was published in Nature Nanotechnology” href=”http://www.nature.com/nnano/journal/v8/n10/abs/nnano.2013.189.html”> Nature Nanotechnology, Cardelli found time to respond to a few questions.
Q: What does this appointment mean to you?
Cardelli: I will have time to think about research issues in a university environment, which is nice. I’ll be in Oxford for about one week a month for the first year.
It also means I can supervise students in an official capacity. Until now, I’ve often been co-supervising students, but I was never the main supervisor because there had to be a university supervisor for that. Now, I can do it in first person.
And I can interact with Oxford departments in particular, where I have people who I already have been working with, both in computer science and in physics. That has to do with my interest in DNA computing.
Q: Take a moment to discuss your research focus over your career and your current research interests.
Cardelli: The first part of my career, I worked on programming languages, particularly the theory of programming languages and type systems for object-oriented languages. That spanned 10 to 15 years, but it culminated with a book I wrote [Martín Abadi, who’s now in Microsoft Research, on object-oriented programming and type theory. That closed the chapter for me.
After that, I wanted to take a rest and do something else. At the time—it was around 1996-’97—the web was coming online, so I started working on distributed computing. Andy Gordon, who soon became my colleague at Microsoft, and I worked on a model of concurrency called the ambient calculus, which is used to describe mobility of our software and devices over a network. That became fairly popular because it was timely, and that was the second phase of my career, until 2002 or so.
At that time, I got a fellowship at the Royal Society, which got me more interested in scientific questions, as opposed to the standard computer-science questions. I started working on physics and biology. I was invited to a DNA conference to give a keynote talk. I learned about a lot of new things happening in DNA computing, and I really cared about that. That is what I’ve been doing since.
Q: What got you excited about that area?
Cardelli: DNA computing started almost 20 years ago with a notion of implementing algorithms using DNA molecules. More recently, the new emphasis is implementing new computational models in DNA and, in general, in some molecular substrate. A computational model could be Turing machines or it could be concurrent models. We’re working on trying to implement chemistry itself.
Chemistry is a collection of chemical reactions, and they describe chemical systems. That’s a descriptive language; it describes what happens in nature. But in DNA computing, we want to use it as a programming language. We want to write programs using chemical reactions, and the class for the algorithms and systems you can describe that way is quite interesting.
You want to compile this language to something molecular. The only way we know how to run it is to compile it to DNA. You can program DNA to do more or less what you want. That’s what we’ve been doing.
Q: How important is effective collaboration between industry and academia?
Cardelli: While it’s very important, in academia, there is a lot of interesting industrial collaboration, because—especially here in the U.K.—the funding atmosphere is becoming very applied, so there’s an interest in getting industrial partners. But it’s important for academia to get this kind of collaboration, because they get access to either data, raw data that they don’t have access to in the academic environment, or problems they would not be exposed to without looking at the industrial environment. That is motivation for academia.
For us, the motivation has always been that we need to engage with the academic environment because that’s where we can exchange and grow ideas. Without that, we would not be able to function as a research organization for industry.
Both sides need the other, so that’s a happy arrangement.
Q: What value do you derive from teaching the next generation?
Cardelli: That’s a quite interesting question, because I’ve been working in an interdisciplinary area, at the border between computing and biology. Nobody is trained in this interdisciplinary area traditionally. When we look for somebody to work with, there are no such students.
There are people either training in math and computer science or training in biology. There are now a number of interdisciplinary master’s degrees, for example, to try to merge these two communities. That’s what’s interesting and exciting about teaching to a new generation. The exciting research we think will happen in the future, at the border of computing and biology, will require a new kind of people, so we need to train these people. We need to engage more in these kinds of interdisciplinary activities, because they will become important areas of activity and of commerce.
What is going to push this is the fact that computing is getting into these areas. Mathematics got into biology many, many years ago, and it was very successful in its own way. Now, computing is getting into biology, and I think it will be very successful in its own way. The effect of computing on the other sciences will drive the need for new kinds of teaching programs and a new kind of student for this century.