MELANIE MOSES — PROFESSOR OF COMPUTER SCIENCE AT UNM
THE BASICS: Melanie E. Moses, 52, born in Washington, D.C.; married to Rosa de la Vega since 1992; one child, Alejandro Moses de la Vega, 18; Ph.D., biology, University of New Mexico, 2005; postdoctoral fellow, scaling in biology and computation, 2005-2006; bachelor’s in symbolic systems, Stanford University, 1993.
POSITIONS: Professor of computer science with secondary appointment in biology department, University of New Mexico; has been at UNM since 2008; external faculty, Santa Fe Institute, since 2012; visiting associate professor, UCLA biomathematics, 2014; visiting associate professor, Universitat Pompeu Fabra, Barcelona, Spain, 2013.
OTHER: Board member of the Computing Research Association Widening Participation and Reboot Catalyst, working to increase computer science degrees among Black, Latina and Native American women.
Melanie Moses is excited — but also worried about — artificial intelligence. She also loves thinking about ant swarms, immune systems, volcano emissions and how to keep robots from bashing into each other.
From those seemingly disparate topics, Moses has built a career as a biologist and computer scientist at the University of New Mexico.
In addition to teaching both disciplines, she conducts research through her Moses Biological Computation lab and is an external faculty member at the Santa Fe Institute.
Moses has become more visible in the past year, writing to newspapers and talking to state lawmakers about the perils and promise of artificial intelligence. She served as an expert for a recently passed bill that will require political ads to include disclaimers if they use AI to generate “materially deceptive media.”
One particular focus for Moses has been “algorithmic justice,” or ensuring that the use of advanced AI does not result in unfair or discriminatory research findings or decisionmaking.
“For example, state government might use algorithms to determine who is eligible for a certain benefit,” she said. “We do use algorithms to decide who is released … before their trial starts. The ones that they (state officials) use are probably the better ones on the market, but there are still all sorts of questions about whether the predictions they’re making are accurate.”
Her lab’s volcano work involves using drones to measure carbon dioxide levels in such places as Iceland, New Guinea and Costa Rica.
And the bashing robots? During a robotics contest, Moses inadvertently learned that when she wore bike reflector bands, the robots would avoid her. Thus the solution to outfit the moving computers with a “magic belt” of reflectors.
Problem solved.
What excites you about AI?
“A huge scientific breakthrough using (advanced) AI was AlphaFold, which was able to predict how proteins fold from the genetic
sequence of those proteins. That can help us design new vaccines, design new drugs, design new medicines for all kinds of things. Doing just one of those at a time was sort of a Ph.D. student’s level of work, so four or five years of work. AI has done this now for billions of proteins. It’s unimaginable that humans could do this without AI.”
And what concerns you about AI?
“These bias questions are a big concern. There are lots and lots of examples of AI taking these past prejudices and projecting them into the future with the idea that there’s no bias there. There’s sort of a human reaction to say, ‘Oh, it came from a computer, it must be objective.’ Sometimes I’m optimistic that at least we’re aware of this problem, and so we’ll fix it. But I don’t know. Some of the facial recognition algorithms, we’ve now known about the biases there for many years. Certain software is fixed … and then other software, it’s just not. And these problems keep reappearing over and over. Sort of whack-a-mole.”
Tell me something interesting you have learned from your research.
“One example of an insight that I think we’ve gotten from this work is that randomness is really important. And this was a surprising result. So I spent a lot of time following ants around in the desert, and when you do that you begin to think the ants are very foolish. We think of ants as laying a pheromone trail, and you leave food out on a picnic blanket and they all come and they take it. But it’s really only a small number. Often, the other ants are off looking for other piles of food, and when you know there’s already a big rich source of food, it seems that’s not a very smart move. But in an unpredictable world, that’s exactly what you want to do. You want to be always looking for the next thing. And in the (human) immune system, there’s a similar sort of dynamic.”
How so?
“When your immune system … has found a virally infected cell, you don’t actually want everything to come to that location, because there might be something else in another location. The lesson the ants taught us for the immune system is that exploration is really important — this random search for things that you don’t know what you’re looking for. You need much more randomness than I think you intuitively think you need. As humans, we tend to want to follow the crowd. And I think that ants are much better at the exploration piece of that, and maybe we could learn a thing or two.”
What’s been a difficult thing for you?
“My undergraduate program (at Stanford) was tough. The hardest class I took was firstorder predicate calculus, a logic class, essentially. We had takehome exams where you had to write these really long proofs. And I remember one, I stayed up all night to work on this thing, I turned it in and I get there, and there’s somebody else’s proofs that were three times as long. I did get through the class, and then we had a reunion and Reid Hoffman (co-founder of LinkedIn) and Marissa Mayer (former CEO of Yahoo!) — these are famous Silicon Valley CEOs — all talked about ‘Oh, God, that class was the hardest.’ It was vindicating. It kind of gives you this sense of “Well, I got through that. I can do hard things.’”
Have you always been interested in science?
“Yes. I usually credit my grandmother, who was a math teacher. A special ed teacher, actually. She was an African-American woman and clearly a genius. My earliest memories are playing math games with her on the telephone. My dad was an engineer, and so I always just loved math and science. I started programming, probably as a junior high student. I found that you made more money coding than you did working at McDonald’s.”
What are your goals?
“I think we’re at a pivotal point with AI, where we need to figure out what do we, as a society, want to use AI for. It’s clear that AI can do great things that benefit lots of people — improving health, food production, clean energy — but it can also do pretty terrible things that will perpetuate our history of bias and discrimination, produce deep fakes and misinformation that erode our already tenuous faith in democracy, concentrating power in a few small companies. Part of what I want to do is contribute to how we shape the future use of AI for human benefit.”