Dr. Thenkurussi “Kesh” Kesavadas shares his vision for the Albany Artificial Intelligence Supercomputing Initiative -- Albany A.I. -- a $200 million public-private supercomputing initiative based out of UAlbany’s soon-to-be-renovated College of Engineering and Applied Sciences (CEAS) building. Buoyed by $75 million in state funding to complete the transformation of the former Albany High School into a state-of-the-art engineering college and fund the construction of a supercomputer, Albany A.I. will significantly expand New York’s capacity in this major emerging field.
More about Dr. Thenkurussi “Kesh” Kesavadas, Vice President for Research and Economic Development, University at Albany
Website for UAlbany's Division for Research and Economic Development
More information on Albany A.I.:
"UAlbany Engineering Building to Anchor New Artificial Intelligence Supercomputing Initiative"
"The Albany Artificial Intelligence Supercomputing Initiative," a message from UAlbany President Havidán Rodríguez and Dr. Kesh Kesavadas
Follow the Division for Research and Economic Development on Twitter-- @ResearchUAlbany
Follow Kesh on Twitter -- @KeshUAlbany
The Engagement Ring is produced by the University at Albany’s Office for Public Engagement.If you have questions or comments, or want to share an idea for an upcoming podcast, email us at firstname.lastname@example.org.
Visit the Office for Public Engagement online.
Transcript of The Engagement Ring, Episode 6: Albany A.I. Supercomputing Initiative
[Lively, upbeat theme music plays as program host Mary Hunt introduces the program and plays excerpts from Dr. Kasavadas’ interview.]
MARY HUNT: Welcome to the Engagement Ring, your connection to an ever-widening network of higher educational professionals, scholars and community partners working to make the world a better place. I'm Mary Hunt. Today on the podcast…”
KESH: What we are trying to do in our approach is to create a computer that democratizes AI. When I say it democratizes AI, I mean that every faculty and student should have access to the kind of power to do AI that industry has access to.
MARY HUNT: I’ll talk with Dr. Thenkurussi Kesavadas, vice president for research and economic development at the University at Albany.Dr. Kesavadas -- Kesh to all those who know him -- is the driving force behind Albany AI, a $200 million dollar public private supercomputing initiative based out UAlbany’s soon-to-be-renovated College of Engineering and Applied Sciences in midtown Albany.
KESH: We are launching into something that, um, is very unique and probably one of the, um, leading AI infrastructures at any university in the country.
MARY HUNT: Albany AI is expected to significantly expand the artificial intelligence supercomputing resources available in New York for teaching and research well beyond traditional STEM applications, including the arts and humanities, and help propel upstate New York to the forefront of next generation chip design.
KESH: We will see a huge impact on this region. I would like to say that it could become the next research triangle for AI, will be in New York State.
MARY HUNT: Here’s my conversation with Kesh…
MARY HUNT: Welcome to the podcast, Kesh.
KESH: Thank you, Mary.
MARY HUNT: And congratulations on the launch of Albany AI. You must be very excited about the prospect of completing construction on the College of Engineering and Applied Sciences and also undertaking this huge initiative multimillion dollar research initiative. Um, it’s a huge undertaking.
KESH: Yeah, indeed. I think we are launching into something that is very unique and probably one of the leading AI infrastructures at any university in the country and, at the same time, building a new college of engineering which is going to be very innovative in nature, also very much better integrated with AI than a traditional engineering program. Yes, lots to do in the in the next few months.
MARY HUNT: Absolutely. I hope you’ve got your vacation behind you.
MARY HUNT: The term AI, or artificial intelligence, conjures up a lot in the ear of the beholder. It really depends on what’s your experience, what’s your knowledge, what’s your exposure to AI. So, could we start with a few definitions starting with artificial intelligence? What does AI mean?
KESH: Yeah, of course. The term AI has been around for very long. From the time science fiction writers have been thinking of artificial intelligence, you know humanoids and computers, and robots that can take over the world, right? So the concept of AI in the Hollywood world has been around for a very long time, but the field of AI from the scientific perspective has also been going on for a very long time. The idea that computers can learn and try to behave like human beings has been like the goal of AI research for many decades now, but it's only more recently that AI systems are being developed or that is getting really what I would call intelligence into it, so we are really in a threshold where this idea of AI, artificial intelligence, becoming a reality is something we can start seeing happening now, and we can see where it can lead to in the future.
MARY HUNT: What are some common examples of things that people might not realize, you know, are examples of AI in practice in their everyday life?
KESH: Okay, so AI is very broad, but maybe I'll give you a few examples. Uh, in the natural language processing, understanding language, for example, AI is an absolute tool that is required to learn from millions of words and combinations of words and things like that. Learning one language and converting into another language, for example, that's a very good example of how AI is changing our understanding of human communication, right. If you look at the other extreme in the scientific domain, AI is playing more and more a crucial role in fields like medicine, in diagnosis right like AI is helping look at X-rays, for instance. AI is helping in drug discoveries, doing precision medicine. Learning from genomics and creating treatments for patients, which is customized for the patient, rather than doing something which is designed for millions of patients at the same time. In between, AI is everywhere. In social media there is AI which is monitoring what you're doing online in Facebook and other social media and creating content and interactions for you that actually is using your emotions and interests in a fashion that perhaps even you don't realize. Even that is a part of AI. So, of course, AI is very broad. I just gave you two or three examples, but I’m sure as we talk maybe there are other things that we can discuss.
MARY HUNT: Yes, back to those definitions -- supercomputer because that figures into Albany AI. What is a supercomputer? And cluster is also used. Um, is that synonymous with supercomputer? What’s the distinction?
KESH: Yeah, yeah, let me address that. The word supercomputer describes, generally speaking, computers that have been designed to solve very large physics problems and so sometimes also called high performance computing, HPC, and so those kind of supercomputers have been there for a few decades as well. But, more recently, as AI is becoming a big part of what we do in looking at computers that can learn from very large data, they're not really designed to solve large equations. Right, it's not like solving an equation to describe the weather. It is learning from millions of words and understanding the meaning. So, for those kinds of problems, we need computers that are fundamentally different in the way it computes inside than a traditional high performance computer. So, when we refer to super… AI supercomputer cluster these are clusters of small computers, which sometimes we call GPUs that can be stringed along to create and solve many large problems that smaller computers may take months and months to do, but these computers could learn in days-time or in minutes-time. So, it’s inherently creating a new generation of supercomputers designed for AI but not for doing traditional scientific exploration.
MARY HUNT: This concept of learning in humans versus AI learning or machine learning -- can you differentiate? We all know how humans learn. How is learning in a machine different? I think this is a place where a lot of people struggle -- trying to imagine how we comfortably and capably are in control of such a powerful technology.When we say learning in terms of AI, what is the difference there? Machines don't learn like humans.
KESH: Yeah, that's a very good question. You're right. Human beings, learn and internalize knowledge in a different way. We also apply our experience in life. We apply emotion, our understanding of the world, apply your knowledge, right. Machines learn in a completely different way. Machines can learn from very large data and do inference and see patterns that sometimes human beings are not capable of doing because we, we tend to… you know, look logic behind what we see and learn in a different way. So, machines can take extremely large data and see correlations that humans may not see, right. So, it's inherently the learning that machines do because of that is very different than what humans can ever perceive to do, right. For example, machines can repeatedly look at data and give a solution which might be very accurate, let's say, if it is trained perfectly well. Human beings, however well we are trained ourselves, we could still make human errors because of multiple reasons. Maybe we are fatigued, or we are in a bad mood, or we have our own internal biases and so on and so forth. Machines can very, very you know, repetitively do something in a very narrow space if it has a very large knowledge put into those computers. So, you can see that AI could be extremely reliable in a certain way, but humans could be very good at making inferences that the AI if it has not been trained cannot make. But more and more what we're seeing is that this correlation between two different things that AI can make, humans can sometimes not see those, because we’re not trained to look at that. And all of us in AI have seen a correlation between weather and health, for example, that human beings, you know don't see it. So it's very exciting to think about what the impact is going to be on human life and society in the years to come, as these AI systems become more and more powerful and better and better at learning these relationships.
MARY HUNT: You’ve said that Albany AI, you think, has the promise to transform advances in a number of fields – healthcare, climate, weather, national security. Is it too early to sort of tell us about what you imagine these transformations could be? What kinds of things are possible?
KESH: Yeah, so let me start by saying that universities have struggled to get access to using very large AI computers, because some of the best AI computers today reside with large companies, right, like social media companies or in the hands of large defense establishments. So, for university researchers to do AI and to solve their problems, very often, they have been building their own clusters, smaller clusters in their labs and things like that. What we're trying to do in our approach is to create a computer that democratizes AI. When I say democratize, I mean that every faculty and student should have access to the kind of power to do AI that industry has access to, right. If you do that, coming back to your question of how do you think it's going to make an impact, what we can see in the future is that now faculty members can start doing really innovative work in the space of AI in, say for example, looking at solutions to help a public health problem; trying to predict a next generation of pandemic by looking at data that human beings cannot see; looking at the effect of weather on the health of the population in an area; looking at how climate change makes a difference. So, there are a number of examples that can help the society for which you need to solve and learn from very large data sets and those kind of things are very hard for faculty members to have access. So now to come back, how do we think our superior will make an impact, I think this will give an opportunity for our faculty members, scientists and students to now start imagining that they can now solve these problems on their own and in a few years’ time, we might see results from the supercomputer helping the society in designing and planning to address climate change, for example.
MARY HUNT: Kesh, how do you see Albany AI as impacting UAlbany’s new and emerging College of Engineering and Applied Sciences?
KESH: Ah, yes, absolutely. So, AI is, in fact one of the fundamentals of the deep technology that you see everywhere. I’ll give you an example: self-driven cars. They use AI to make it go automatically on the road. For example, AI’s being used in drone technologies in different kinds of areas like in urban planning and in smart health and things like that. Our engineering program, which is very new, right now has a focus on computer engineering and computer science and we are also envisioning that in the very near future we'll have areas like mechanical engineering. These programs will be designed to create workforce or engineers who can work on the next generation of robots, and develop the next generation of cars that will be safer, can go anywhere without you driving, and so on and so forth. Maybe in designing the next generation of medical devices that can treat people, and so and so forth. Ah, there is also huge interest in fields like virtual reality and augmented reality. That's the future -- building metaverses where people interact with each other in the virtual world, and AI plays a very big role. So, I think that our new engineering program will be actually leveraging these things, perhaps even teaching classes in metaverse, students working together to design next generation robots. So we think of this AI initiative playing a very crucial role in how we design the next curriculum and how we train the students of the future.
MARY HUNT: And the timing seems to be spot on. I mean just recently Congress passed the Chips and Science Act of 2022. So, obviously there is a need, and hopefully will be funding and support for programs like this. How do you expect that that act may affect the work that we do here at UAlbany in terms of artificial intelligence? Is there a connection? Will there be opportunity through that?
KESH: Yes. So, Albany, as we all know, is already a leader in the semiconductor field -- Albany Nanotech, NY CREATES, and CNSE and others -- you know, one of the most advanced places to do this research in the U.S., if not in the world. The new CHIPS Act, if you look at the whole bill itself, it’s a very high tech bill in the sense that it’s focusing on making our country very competitive by innovating in chip manufacturing, making chips available internally. But it's also focused on investment in the area of robotics, drones and health IT. So, if you look at all the things, our expertise at UAlbany in the space of weather and so on and so forth, will directly leverage this, but more interestingly in Albany we have been focusing on nanotechnology and manufacturing and fabrication of chips. The AI initiative is looking at how can we build an AI supercomputer and create applications using AI. So if you combine these two, we as a region, the Capital Region, and UAlbany leading that, could actually be a place where you come to design a chip, manufacturer a chip and build a computer that will solve the challenges of a society. I will actually venture as far to say that we will be uniquely positioned in the U.S to do everything from end to end when the chip is considered. Because of that I think the CHIPS Act is really exciting for us because it actually provides impetus and hopefully it will lead to innovations that that you know rest of the country can benefit from.
MARY HUNT: What is the global picture when it comes to AI? Who are the leaders in AI? Is the United States among those or is it in the lead position? Tell me a little bit about the competition for AI?
KESH: Well, that's a very tricky question. Anybody who reads and follows the field of AI knows that there are other countries who have invested more in the space of AI. In fact, China is an example of where a tremendous investment has been made in the AI space. And for us to retain our leadership we need to be investing at the same level. And that's why the CHIPS Act is so exciting for everybody involved is because this is the first time that our country has come together -- industry, federal government, research institutions -- to address the issue that we need significant investment to be in the lead, or else we will… I would call it you semiconductor arms race is going on there, we will lose this battle, and with this I think we will recoup in the next few years, next two, three years, you will see that we as a country will retain our lead and continue the thing that we do the best -- that is innovate and innovate and innovate.
MARY HUNT: Who are your partners in the AI initiative?
KESH: We have very good partnerships with the industry, that already work with the SUNY system through NY Creates and it includes all the companies you know who work in the space of AI; IBM comes to my mind, is a company you know with whom we are working in terms of understanding the vision, this field and so on and so forth. We are also writing grants to the federal government to create institutes in AI here, other federal government agencies like NOAA, Department of Energy, Department of Homeland Security, and others who need AI for their own areas of research, so we are working with them as well, to create the federal partnership, industry partnership, and we’re also working with more organizations in this region, ah, in medical healthcare systems, New York State government looking at how government technology can benefit in the space of AI. So, I would say that the collaboration is very broad right. And we can’t forget other institutions, other SUNY institutions and other colleges in this area who are also partners, like RPI and HVCC. This is not a field that is small; everyone has to come together, right, and that's the kind of thing that we are trying to build here.
MARY HUNT: And UAlbany has already established a record in terms of AI. Can you talk a little bit about some of the key projects related to AI that are going on across the disciplines at the university? Weather and climate, come to mind.
KESH: Yes, yes, obviously, since you brought up weather and climate, yes. We have a very strong group in the space or atmospheric research and weather. They're using AI like, for example, monitoring the New York State Thruway through cameras, looking at the weather prediction for snow, for example. Looking at AI coming from data that we collect through Mesonet. That’s a network across the state that is managed through our university’s Atmospheric Sciences Research Center where they get data from all over the state. You can use that to do predictive modeling. We also have a National Science Foundation AI Institute for Trustworthy AI in collaboration with Oklahoma University here. So, those are some areas of weather. But, if you look broadly at campus, there are many pockets of excellent work that’s going on in the space, ah, chemistry modeling and even in philosophy, mathematics looking at how do you trust AI, what are the ethics of AI, how do you monitor environment through cameras, what is the ethics behind that. So, I would say that there are a surprisingly large number of people on campus and faculty members, scientists working in this space. With this initiative, I think we're going to see this mushroom.
MARY HUNT: One of the concerns people express about AI, rightfully or wrongfully, is that it could eliminate jobs. Is that true or will AI actually create jobs, maybe different jobs? What is the real truth on that?
KESH: Ah, this is a… I mean it’s a question which we have always asked for the last 200 years. You know, every time there's a new technology coming, we are worried that the jobs will disappear. It happened when the steam engine came, and when the industrial revolution came. It happened when computers became prevalent. But those kinds of fears have never materialized, simply because every time you create a new technology or new revolutionary concept you're creating a whole generation of new workforce that's needed to build that area, like, for example, let’s say take AI. Self-driven cars, that’s the future. They drive themselves. There are hundreds and thousands of engineers and scientists who sit and design the AI behind that. It created a whole class of jobs that did not exist 10 years back. Same things, take healthcare. Well, you're now seeing software development happening, creating jobs, or creating AI. So, I don't think the jobs will ever disappear. It moves from one sector to another sector, right. I think the bigger fear that people have is losing their job the way they do it today. It's not because jobs don't exist. So, I think that's why it is important that we think of this as a state and as a country providing new pathways for people to up-skill, change their skills, right, so they are prepared to go and work in a slightly different industry, but nevertheless jobs are not disappearing. I think more jobs are being created every day because of AI.
MARY HUNT: Albany AI includes something called Albany AI Academy. What is that?
KESH: So, the concept of AI Academy is something that we are very seriously planning and envisioning, and that is introducing AI to all the students who come to our campus, okay. And not just that. Also creating programs that will address some of the things I mentioned earlier, that is, how do you up-skill, how do you train the workforce, how do you help people change their career, which means that these are programs, microcredentials available for the public, creating programs for industry so that you can train lots of people. Our AI Academy concept is that we will address everything from education for undergraduates all the way to professional development, which will be a service that we provide for society.
MARY HUNT: So you have developed a concept of something called AI plus X. Can you talk a little bit about what AI plus X is and how you envision that being rolled out?
KESH: Yes, so the idea of AI plus X is that you could come to University at Albany to do any… to learn any subject, any subject -- let’s call it “X” – and you will also get education in AI. That's the idea. We call it the AI plus X or some people prefer calling X plus AI. You do X plus AI. So, the big vision picture that our provost is leading and all our deans are working on, is to create a program where you can come to Albany to do athletics or zoology or music, but from day one when you come to Albany, you’ll be introduced to the fundamentals of AI, what AI means, how AI can be used, what are the ethics and the do's and don'ts about AI, and then specialize as you go along and as you become a senior in your final year to learn more about AI in your own domain.For example, if you're doing music, you now learn about how AI plays a role in music. So this is a campus-wide initiative to fundamentally, you know change the way students understand AI from the perspective and not learn about that from the Internet, but get foundational understanding about AI. So, it's a very exciting initiative, and I think that, uh, I'm thinking that many universities will see what we’re doing at the University at Albany and try to implement something similar to this.
MARY HUNT: Why is that important? Why is it important that they get an introduction to AI in whatever their chosen discipline is?
KESH: You know I like saying that AI is the new Humanities, right. In any curriculum in the country students learn about humanities in any subject they learn, right. AI is almost becoming like that simply because AI is so prevalent in everything. To be the workforce of the next generation… If you don't understand what AI is then you're not prepared for what is going to come in the future, and that is important to know because if AI is doing everything for you -- making decisions, it’s monitoring people’s safety and security -- we want domain experts who work in the industry to understand it, so they can contribute to the design of next generation AI, right, not just from the scientific domain, but also from demography, from where they come, from their ethnicity. Everyone has to contribute, or else AI will be designed by engineers who do not look at the society as a whole, but who may be looking at only a part of the society. It works very well for some people, but doesn’t work for other people. We don't want our students to be like that when they go. We want them to go and contribute to this, so that when the next generation of AI is being designed, they can provide their input. And so, we’ll have AI which is more democratic, or more equitable to everybody,
MARY HUNT: When I hear you speak of this I think of the phrase “AI bias,” which I didn't really know what it meant, so if you'll just take a second to explain what AI bias is and how you can alter that or affect what is AI bias.
KESH: Yes, so AI bias is a real thing. One good example of that is software that does facial recognition, right. Those softwares are known to have a racial bias, or they very often don't do a good job when the people whose face is being recognized come from different ethnicity, right. Those kind of biases may not have been intentionally designed but the way the AI systems have been trained, those biases were built into it, right. And if you take that and think more broadly -- think of AI in healthcare right… if there is an AI system which has a bias, it may impact the diagnosis that the computers will do for people, based on race and things like that. It may not actually miss all the health problems for a certain segment of people, but some other segment it may not understand because the AI didn't quite understand the effect or impact of ethnicity or where you come from, or things like that. So, biases in AI do exist, and one way to remove the bias is having more input from a broader set of society, so that AI tends to become much more reasonable or fair.
MARY HUNT: Kesh, what are you anticipating the effect of Albany AI to be on not just the Capital Region, but on New York State?
KESH: Yeah, so you know, as I mentioned earlier that the impact of the semiconductor industry on this region of the State of New York has been tremendous, right. It's attracted large companies like GlobalFoundries to come here and so on and so forth. What my vision is that the Albany AI Initiative will have a similar kind of impact in the State of New York in applying AI in every different field, which means attracting companies to move to this region, or move to New York, because this is where the talent exists. This is where some of the best computers are available for people to design. My vision is that we not only create the best students for the next generation workforce, but this will also attract the economy to grow in the space of AI, create partnerships that can help in bringing larger institutes, centers of learning, industrial innovation centers, so on and so forth to Albany or to the State of New York, which allows access to AI so we will see a huge impact on this region. I would like to say that It could become the next Research Triangle for AI, will be New York State.
MARY HUNT: What use or application of AI do you hope you'll see in your lifetime?
KESH: Well, I have my own biases.
KESH: You know I think that AI will make transportation more efficient and safer. The reason I'm saying that is that, even today, with all the safety features, um, people die in accidents every day. We could see the transportation systems of the future being much more streamlined and safer because of built-in AI and maybe a time will come when an accidental death on a road may be considered as unbelievable. How could such a thing happen, right? But we can see that happening right, I mean, like in the old days in factories people died. Today’s factories are so safe, right because we have all these new rules in the same way, you can see, the transportation becoming safer. Ah, the second thing I think that, um you know that by using the power of AI we might start seeing impact in the way the environment is managed, right. There could be a much more efficient way of looking at the climate changes and making decisions. We know smart cities, for example, that can, you know, um help for example an aging person. Systems that automatically adapt for people who are aging can help people with cognitive issues, making the quality of life better for people who are underserved today. Not just the aging population, but even younger generation people who have disabilities very often are underserved in the sense that there are not many things for them. They’re stuck inside the home. Maybe these AI systems with assistance and robotics can give new life to people who today who have no way but may be stuck at home, maybe due to a vision problem, mobility issues, and so on and so forth. So, I think the quality of life may have a big impact in the years to come through AI.
MARY HUNT: Where does your, I guess passion is a good word for it, because you speak with passion about AI… Where does this deep passion for AI come from? Have you always been interested, even as a young boy, in science? I don’t know if you’re like the rest of us who grew up a generation where we all watched TV and movies and we saw these powerful discoveries and science fiction stories about what was possible that we only dreamed of. Where does this come from?
KESH: Yeah, I used to be a big fan of reading Sci Fi books, you know Isaac Asimov. I always wanted to be a robotics person. I started working on my PhD in Robotics and I got introduced to AI in my first class that I took as a graduate student. The very first class that I took was a course on… they called it Expert Systems at that time. It was really AI neural networks. Ah, so when I was completing my PhD dissertation, I applied AI for robots to understand objects for manufacturing. So, this was in the early 90s. Ever since, I’ve been very interested in AI. And I’ve not only seen this field change, but I’ve also been a part of this change in some sense, you know, using robots for training surgeons, looking at how AI can help in diagnosing a tumor, you know, things like that which you know the area I have myself worked as a researcher. My lab has worked on that. So, it has been a lifelong mission for me to see amazing discoveries happening through AI.
[Theme music plays under the remainder of the program]
KESH: What I would like to say is that coming to Albany and having this AI initiative is like a dream come true for me because now we are doing something at a scale that I did not think we’d be able to do, or I will be able to do, for 10 or 15 years. And now we are actually building those things. So, it’s really exciting, and you know asking about movies… Matrix has been always my favorite movie, you know.
KESH: And I always think that, you know, Hollywood is always several decades ahead of us… they always envision things for the future and we try to replicate that. We say metaverse, well Matrix is a metaverse in some sense, right.
MARY HUNT: Well, you’ve got them beat because being part of the real deal and being there at the forefront of making these things happen and really make changes in people’s lives is much better than making a movie, don’t you think?
KESH: I think so, yeah, I think so. You know, it’s not just entertainment. It’s real life.
MARY HUNT: Kesh, I wish you the best of luck with Albany AI and we’ll be watching with great interest as it develops. Best of luck and thanks so much for being my guest today.
KESH: Thank you very much, Mary. Great questions and I also hope that this will be a big success for the university. Thank you!
MARY HUNT: Dr. Thenkurussi “Kesh” Kasavadas is the vice president for research and economic development at the University at Albany.Learn more about UAlbany’s Division for Research by visiting albany.edu/research. Follow Dr. Kasavadas on twitter @keshualbany.
MARY HUNT: The Engagement Ring is produced by the University at Albany’s Office for Public Engagement. If you have questions or comments or want to share an idea for an upcoming podcast, email us at email@example.com.
[Music fades out]