Interview with Dr. Michael Balazs on Generation AI Nexus

Presenting awards at the November 2018 Generation AI Catch the Phish challenge: Sponsor Kelvin Wright, President/CEO of Cipher Logix (far left), Gen AI lead Michael Balazs (third from left), challenge designer Eric Harley (second from right), and technical lead Reggie Carey (right) congratulate the White Shark Team for their k-nearest-neighbors analysis. Students represented William and Mary, University of Maryland, and University of Maryland University College. Photo: Eny Hathaway

Interviewer: Cameron Boozarjomehri

Welcome to the second installment of the Knowledge-Driven Podcast. In this series, Software Systems Engineer Cameron Boozarjomehri interviews technical leaders at MITRE who have made knowledge sharing and collaboration an integral part of their practice. 

Michael Balazs is deeply involved in a challenge to shape America’s workforce of the future and the economy in which we operate. While helping run a hackathon about identifying possible pandemics, he and Jay Crossler, MITRE’s chief engineer for Learning Systems, began talking about a looming problem.

According to McKinsey & Company, there will be a shortfall of up to 250,000 data scientists in the United States. in a decade. This is troubling given how heavily China and other countries are investing in artificial intelligence (AI) expertise and claims that “the nation that leads in AI will be the ruler of the world.”

Balazs and Crossler discussed this problem with MITRE senior vice president Richard Byrne, and developed a plan to bring AI tools and data sets to students across the country. “We believe if we give students secure access to cutting-edge AI tools and datasets they’d normally never see, they won’t just become engaged with the technology. They’ll deliver real solutions.”

The result of those conversations is called Generation AI Nexus. MITRE launched Gen AI at a hackathon in November 2018, with students from the District of Columbia, Maryland, and Virginia. This spring, the Gen AI team is working with three Early Adopter universities—George Mason, University of Massachusetts Lowell, and University of Texas at San Antonio—to incorporate lessons using computational notebooks into their spring courses.

I spoke with him about the power and pitfalls of artificial intelligence, why MITRE is the right organization to work with students (of any age in any discipline), what it means to build an AI economy, and what human-machine teaming really entails.

Click below to listen to podcast:

 

Podcast Transcript
Boozarjomehri: Hello. My name is Cameron Boozarjomehri, and I am interviewing Michael Balazs for the MITRE Knowledge-Driven Podcast. Michael, do you want to go ahead and tell us a little about yourself?
Balazs: Sure. My name is Michael Balazs as Cameron said. I have been with MITRE now for coming up on eight years. I’ve pretty much spent my entire career down in Charlottesville. I kind of came on to do some specific work, but it took a while to get my clearances, so I ended up getting into robotics. We kind of played around with that for a while, and we got this crazy idea about 3D printing drones and that kind of took off. I keep expecting that work to die, but every year it’s grown. We’re now in our seventh year of doing that.
Balazs: That’s not the only work I do. I like to branch out, I like to do a lot of different things. I’ve been very involved in the MITRE Innovation Program here, I’ve had a MIP pretty much every year, so I love research. Coming out of an academic role, it’s not surprising.
Balazs: I’ve also done a lot with things like non-traditional compute clusters. We’ve built a computing cluster out of web browsers. We just had a URL. Now we’re kind of expanding into the world of AI but not really in the traditional way. We’re really focused on building the AI economy, transforming the US workforce to leverage AI and understand its power and pitfalls. We’re not really trying to create an army of coding monkeys that all know how to do AI, but we’re trying to make sure everyone understands the power and the pitfalls.
Boozarjomehri: Before we get ahead of ourselves, that’s actually exactly what I wanted to talk to you about today. If I recall, you are working on the Generation AI Nexus project or that is basically your baby?
Balazs: Yeah. Generation AI is the initiative, and the Nexus is what we’re calling the technology or the portal that brings all of it together to make it happen. Yeah, it all started as an idea during a hack-a-thon that we did back in April on disease prevention and understanding the spread of diseases and using social media and AI to detect whether perhaps a pandemic was perhaps unfolding in, say, Africa or something like that.
Balazs: The idea came, “This is really neat. We’re exposing students that don’t normally have exposure to AI and they’re really seeing the power of it.” They may not be the experts in it, they may not know how to provide and, say, program their own agent-based models or their own classifiers or anything like that, but they’re seeing the power AI can be assigned to their field.
Balazs: The question became how do we expand that to every discipline? How do we make it so it impacts the entire workforce? That’s really where Generation AI was born out of, and that’s what we’re trying to do. We want to expose every student independent of their field of study to the power of AI multiple times in their academic career.
Boozarjomehri: That’s something I was very curious about is that generation word. If you want to fill in exactly what a generation means in this context? Where is the target demographic?
Balazs: Yeah. The word generation generally makes people think about particular demographics or particular age groups but one of the key things about this is we really want to hit really every generation out there.
Balazs: There’s an article that’s actually coming out here shortly that’s entitled “Generation AI: Not Just For Gen Z”. That’s really a good capture of that. Our initial focus is very much on the next generation workforce, so we’re targeting those who are still in university and the community colleges and the technical schools that haven’t hit the workforce yet. That’s where we’re starting. That’s where the Gen Z focus comes from.
Balazs: We plan to expand down to K-12. We plan to expand into the areas of the mid and late career folks that are looking to leverage AI in their fields as well. This is very much a concern with current government workers that they have all this technology that they’re being exposed to, but they don’t quite understand the power of it, they don’t understand the pitfalls of it.
Balazs: Jim Cook, our vice president for strategic partnerships, has challenged us to really think about how do we not just hit Gen Z but we also hit those folks that are already in the workforce, particularly those government workers so that they also understand the power and pitfalls of AI.
Boozarjomehri: I think that’s something that I had heard about was that this project goes beyond just STEM fields. We don’t want to just be engineers and scientists. We want academics, we want artists, we want historians, we want anyone that thinks they can come up with a good enough reason to leverage AI.
Balazs: Absolutely. The goal here is, as I said a few moments ago, it’s every discipline independent of its connection to AI. I’ve got a friend who is a historian down at the University of Virginia, and he works on Indian and Pakistani relationships. He’s got a treasure trove of documents. He wants to be able to show students how you could go through and use a classifier to do a quick search of those documents to pull out key pieces of information.
Balazs: We’ve talked about how at some other universities, there’s archeology professors that are using AI to do artifact identification. Those are particular niches, but how do we do that across the board? How do we expose all historians, economic students, liberal arts students, social science students, fine arts students?
Balazs: You look at what Google is doing now with computational photography, and it’s mind-blowing. I have one of the Google phones and I can take a picture now at night that has got equal clarity of a $5000 camera. It can do it in a matter of seconds as opposed to a long exposure. It’s that that we want to show the power of and bring it to every field.
Boozarjomehri: Actually that’s really interesting because of the fact that I also have a different Android phone but because of the way software works I was able to download Google’s same suite, and even though I don’t have the same hardware I’m running the same software. I’m still getting the benefit of those nighttime photos. That is a taste of the power that this level of software, this level of development can really bring to everyone regardless of the platform.
Boozarjomehri: Actually, in that point, I kind of wanted to expand on the different age groups since you mentioned that this goes beyond just students and some people who are in late career and government. The point is to target everyone.
Balazs: Eventually. Within reason.
Boozarjomehri: That goes beyond just government. That means if I am a secretary or if I am a mechanic or someone who is trying to retool themselves for the emerging workforce that’s kind of where you’re hoping that one day those people will also be part of this cohort.
Balazs: Absolutely. There’s a lot of discussion right now about exactly how much automation and in turn AI is going to affect the workforce, how much job displacement is there going to be, how much of a job shortage is there going to be in folks who have those skills?
Balazs: Our focus is really about that human/machine teaming side, showing the power of humans working together with AI and automation. I’m stealing this from a friend of mine, but letting humans do what humans do best and letting machines do what machines do best.
Balazs: How do we get those routine things taken care of so that humans can really push the limits of what they’re capable of independent of their job? Independent of their being executive assistant or a mechanic? How much can we advance all of those fields in our economy with the application of AI?
Boozarjomehri: I understand that when it comes to human/machine teaming, there’s always this concern that we’re going to just give the machine a lot more of the work to the point where bias can be introduced or things can just go wrong without us really appreciating it and that becomes very difficult for us to undo that training and in turn it can come to harm … Depending on what that system does, it can come to harm populations or come to have real adverse outcomes compared to what the original intent of an algorithm or tool was. How are you trying to account for that bias and those kinds of issues?
Balazs: We’ll look a little bit at the data side, but that’s not the main focus of what we’re trying to do. I think one of the best ways to combat that problem is people need to understand what the technology is doing and how it works.
Balazs: If I go into a field and I don’t know anything about AI, and I just blindly trust the machine? Yes. We are absolutely going to have these problems. You’re going to have bias, you’re going to have these results that could be very negative to certain populations or certain groups or lead things in completely wrong directions. You can see that in even archeology, misidentifying things.
Balazs: If you imagine the fact that we can expose people multiple times during their academic career, everywhere from K through 20, multiple times to not just the power of AI but the pitfalls and they begin to think about the fact that, “Okay, look, I can’t completely trust this. How good is my data? How good is my input? Do I understand where this algorithm came from? Was this algorithm created by a company that has a bias that might be pushing a certain something? Was it produced by academia?”
Balazs: All of those questions and I can’t possibly enumerate all of them, but the more we expose people to both the power and pitfalls of any given technology the more they’re going to be apt to question it and reduce those kinds of problems.
Balazs: When society pushes back that’s when we solve these types of problems. You know? If people blindly follow things, yes, we’re going to have problems. In the US we don’t tend to have that problem. We’re questioners, we’re challengers. You know, that American spirit that we have.
Boozarjomehri: That’s an excellent way to put it. We’ve seen a lot of technologies. They seem very scary at the outset but as people get that exposure and that context, they come to realize not only is it not something to be feared but it’s something they can work with government or with these groups to make better and serve everyone.
Balazs: Exactly. You’ve got to remember that technology in and of itself is not good or bad. Technology is technology. It’s how any individual person uses it, how much any individual person relies on it. It’s on us, not the technology, to solve the problems of bias.
Boozarjomehri: I think my last real question for you is why MITRE? Why does this organization … What do you feel that we are the place where this should be starting? Where this idea should blossom and go out into the world? Why not maybe industry or commercial groups that typically we’ve seen fill these gaps?
Balazs: Sure. Let me answer that second part first and I’ll come back to the bigger question. What we have found with industry in general is that they’re really good at supporting education. The challenge is that they’re really good at supporting education for their own needs. They don’t tend to take the holistic approach.
Balazs: Now that’s not always the case. There’s usually great benefits to the work that they do, but company X will often train you on their technology and company Y will train you on their technology and while that’s in the general sense very good, if you look at what they’re driving and what they’re trying to accomplish a couple of things.
Balazs: One, they’re interested in their own technology being adopted. Two, they’re not looking at the full spectrum. A lot of these companies are addressing their current needs and their current needs are talent gaps in folks that understand AI but it’s in that deep AI, the people that can come in and really advance their AI capabilities, which is not necessarily what we’re trying to focus on.
Balazs: It’s a part of it. We’re trying to hit that broad part that we talked about earlier. Your average industry tech company is not trying to educate historians on how to use AI. The MITRE premise is that if you really want to shift the entire economy, not just an industry, which is what industry is focused on, if we want to shift the entire economy, we need to educate everyone who is going to be part of that economy. That’s where the MITRE focus comes in or that’s where the MITRE difference comes in.
Balazs: That’s also why MITRE is tackling this. It’s that holistic view. It’s for the betterment of the entire nation view. We don’t want to lock into a very specific particular industry partner or the like. We really think that by taking this holistic view that’s how we are going to play a part in shifting the economy to be this AI economy of the future.
Boozarjomehri: Thinking about the why MITRE, what would you like to expand on?
Balazs: Sure. Whenever I talk about MITRE and the unique role that we play … You know, we’re not allowed to compete with industry, and we play this very special role on behalf of the US government. What I like to tell people is we work in three main areas.
Balazs: We like to tackle the problems that no one else has been able to solve, those really hard problems. We like to tackle those problems that there’s been no profit motive in or maybe industry might be biased around. MITRE comes in in our non-biased, our not-for-profit role to tackle. The third is we look at what technologies 5, 10, 20 years down the road are going to be game-changing.
Balazs: I really feel that the Gen AI project hits on all three of those. This is a very, very tough problem that no one is looking at from the holistic perspective. Industry is looking at it in each of their own individual niches, which is great and that’s definitely a part of the solution, but MITRE can come in and look at it the holistic perspective. That kind of actually hits on both of those first two.
Balazs: Then the third one, we really feel that five, 10, 20 years down the road as AI becomes more powerful and we move towards this idea of generalized AI as opposed to narrow AI the government needs to understand how that’s going to play a difference.
Balazs: As the workforce goes out to rely on AI, then MITRE will have done its job of making sure that the world is safe and the US is leading in these areas as they participate in Gen AI in their education programs.
Boozarjomehri: Very well put. Do you have any closing comments or anything else that you think people should know about the Generation AI project or that you’d like to share with us?
Balazs: Let me just close with an analogy that I often use when I’m presenting Generation AI. I’m a product of the ’90s education system, and as I was going through the education system we were always required to take a computer lab class. This is separate from, say, a computer science-focused class but a computer lab class.
Balazs: As I was going through, we started with something like logo deterral and eventually learned typing. What was interesting is that by the time I finished my educational pursuits before college it was all about the web and the internet and those types of technologies.
Balazs: What became interesting is that when my generation entered the workforce, they demanded access to those technologies. They were dependent on those technologies. They transformed what the US economy looked like. At the end of the ’90s you had a lot of big tech companies, and then you had the dot com boon, and you had less tech companies.
Balazs: It wasn’t really until that generation, my generation, entered the workforce that you saw that transition, that transformational effect on the economy where every company, the mom and pops, the Walmarts, and all of the companies entered into having the web as a central part of their business. We’re trying to replicate that same thing around AI. If we expose everyone to AI now as part of their education program, they see that as another tool in their toolbox when they enter the workforce, and we can see that same transformational change as we did with the internet with AI.
Boozarjomehri: Thank you so much for your time.
Balazs: Thank you very much.
Boozarjomehri: This is my conversation with Michael Balazs about the Generation AI Nexus Project. Thanks.
Balazs: Thank you.

Cameron Boozarjomehri is a Software Engineer and a member of MITRE’s Privacy Capability. His passion is exploring the applications and implications of emerging technologies and finding new ways to make those technologies accessible to the public.

© 2019 The MITRE Corporation. All rights reserved. Approved for public release.  Distribution unlimited. (Case number PR_18-1985-12)

MITRE’s mission-driven team is dedicated to solving problems for a safer world. Learn more about MITRE.

See also: 

Is This a Wolf? Understanding Bias in Machine Learning

Just in Time Science

Building Smarter Machines by Getting Smarter About the Brain

The World as It Will Be: Workforce Development Within and Beyond MITRE

Catch You Later: Recap of the Generation AI Cyber Challenge

Phish, Flags, and Lesson Plans: Upcoming Hackathon for Generation AI Nexus

Technical Challenges in Data Science

Defining, Applying, and Coordinating Data Science at MITRE

Upgrading Machine Learning. Install Brain (Y/N)?

Archives

Pin It on Pinterest

Share This