AI for Social Good: Venu Govindaraju and UB's Expanding Vision

n this episode, I speak with Dr. Venu Govindaraju, SUNY Distinguished Professor at the University at Buffalo, Senior Vice President for research, innovation and economic development, and founding director of the National AI Institute for Exceptional Education. We discuss the Institute’s continued progress in supporting children with speech and language processing challenges through innovations such as the AI Screener and AI Orchestrator, along with the broader work of UB’s research in AI.

The conversation also explores UB’s growing leadership in artificial intelligence, the promise of Empire AI, and what these developments could mean for education, workforce development, and Western New York. This episode highlights how AI can be used for social good to improve outcomes for children, families, educators, and schools.

Find us on Twitter @NiagaraErie

Or visit our website E-NSSA.org

Transcript

SPEAKER_01 0:04

Hello, everyone, and welcome to the Eerie Niagara School Superintendents Association's podcast, achieving joy and mastery in public schools. I am your host, Brian Graham, superintendent of the Grand Island Central School District. We are really excited that you're listening today. This podcast is designed to celebrate all that is good in public education around Western New York. We will be featuring programs and innovative ideas that inspire and influence our students, faculty, staff, and community in new and exciting ways. So, everybody, let's get started. Today on the podcast, we welcome Dr. Venu Govindaraju, SUNY Distinguished Professor at the University of Buffalo. He is the senior vice president for research, innovation, and economic development, and he's the founding director of the National Artificial Intelligence Institute for Exceptional Education. Dr. Govindaraju is a nationally recognized leader in artificial intelligence, research, and innovation, and his work continues to place the University of Buffalo at the forefront of using AI for meaningful human impact. In this episode, we revisit the important and evolving work taking place at UB and explore how the National AI Institute for Exceptional Education continues to advance its mission to support children with speech, language, reading, and writing challenges. We discuss the progress of the AI screener and the AI orchestrator, as well as several other emerging tools designed to help families, educators, speech language pathologists, and specialists in our schools respond earlier and more efficiently to student needs. This conversation will also highlight a major theme emerging from UB's recent research and their publications. Artificial intelligence is not being developed here as a novelty or a replacement for human expertise, but as a practical, human-centered tool for social good. We explore how the institute is working to extend the reach of professionals in the face of national shortages in speech and language pathology and the related services connected to that, while also keeping trust, transparency, and responsible design at the center of this work. We will also take a broader look at UB's expanding role in AI across Western New York and, of course, the state, and that includes its leadership in research and innovation, its interdisciplinary vision for AI education and workforce development, and the growing opportunities connected to Empire AI and the expanded computing capacity that's being helped and funded by New York State. Dr. Govind DeRaju shares why this moment matters, how Buffalo is positioning itself for the next wave of AI-enabled discovery, and what educators and school leaders should be paying attention to right now. This episode is especially relevant for educators, families, and leaders who want to better understand how artificial intelligence can improve outcomes for children and support the professionals who serve them. It is a conversation about innovation, and yes, that is so true, but also about access and early intervention and responsible guardrails and the future of education in a rapidly changing world. Venu, it is so great to welcome you back. We've done a few of these podcasts already, and I'm just so glad uh to be here at UB with you. So welcome back to the podcast.

SPEAKER_00 4:01

Brian, it's just wonderful to be back. I've been enjoying these conversations that you have had uh over the last few years, and it's good to just check and see, you know, how the progress is. So welcome, you know, to have this conversation with you.

SPEAKER_01 4:15

I am just I'm I gush. I gush when I'm here at UB, I gush being part of the advisory board for the AI Institute. I am so excited about what UB is doing, leading the nation and maybe the world in this work. And it really, in my opinion, has positioned the University of Buffalo in in a greater light in academia and in the world of AI and social good. I'm just thrilled. I met you about three or four years ago, and you invited me to retreat. It was a three-day retreat here at UB, and I soaked it all up and had such a great time. And that was because of that grant, right? The grant that you applied for. And when you did that work, maybe our listeners would love to hear that. You when you and your team put together this grant with the National Science Foundation, that was that was put together pre-Chat GPT, I think.

SPEAKER_00 5:10

No, absolutely. And that's a good point you make, Brian. It's been a few years, but uh seems also like a lifetime if you look at the progress that artificial intelligence has made over the last few years. It has completely permeated every aspect of our life, right? Absolutely true what you just said. You know, when we wrote the proposal, you know, for this award, uh, this must be in uh 2022, and we we won this. At that time, uh Chat GPT was not uh part of our vocabulary.

SPEAKER_01 5:41

I know that is amazing. So and and I gotta say, what I've learned about you is that you were an international leader in artificial intelligence way before ChatGPT. And when I say AI, I mean AI and machine learning, your background with the post offices right across the United States, I think Great Britain, maybe United Kingdom, and Australia, if I have that correct, your work, right, was a cornerstone in the post post offices being able to use your work to scan and understand handwritten envelopes. Is that correct?

SPEAKER_00 6:23

That's absolutely true. And it is quite interesting. Uh, you know, going a few decades back, actually. So, you know, in the late 80s and early 90s, at the University of Buffalo, our team of students and researchers actually developed a system that could automatically read handwriting on postal envelopes. And uh we did take that technology to Canada, uh, UK, Australia, and of course in the US. And in those days, you know, it was uh not uh directly referred to as artificial intelligence. Although, if you come to think of this, you know, a machine is reading something written on an envelope and taking the right decisions as to where the destination is. In those days, uh, they would call it uh computer vision or pattern recognition, which all you know comes under the umbrella of uh AI. Yeah. But yes, you know, we were the leaders, and you made a good point also in your remarks, is that uh at UB, you know, we have been the leaders. You know, we are not just simply following uh the trends. Uh we have had this rich legacy, and uh we continue to lead the way uh you mentioned the project on education, actually, very rightly. So there are many metrics by which you could say a particular program at a particular university, how does it rank? Uh one way to look at it would be, I will argue, is to look at how many federally competitive grants in that space the university has. And at UB, we have two very large competitive grants in this space. You are involved in both of them. Yeah. You're also on our advisory board uh for one of those. So that makes me feel proud. And we are happy to say that when it comes to AI and education, UB is truly leading the way.

SPEAKER_01 8:15

Yeah, there's no doubt. And again, a few years ago, and and for our listeners, before ChatGPT became a household name, this university with uh Venu's leadership applied for a grant to solve a problem supporting speech and language teachers and the work that they do with infants and well, toddlers and young children. And they were awarded$20 million, 10, I think, from the National Science Foundation, 10 from the Institute of Education Sciences. And since then, they've been working on an artificial intelligence screener, and then parallel to that, an orchestrator. I don't know, Venue, if you just want to give just a little bit of an overview. I know some people who have listened to this podcast might have an understanding, but we have new listeners as well. Maybe just share a little bit of information about that particular project and how it's accelerated since you know since you started.

SPEAKER_00 9:11

Yeah, sure thing. You know, so let's call it pre-po uh pre-Chat GPT era when the proposal was written. So the idea was to come up with artificial intelligence tools to do two things. First, develop a screener, and the idea is that you should be automatically able to tell whether a child uh is having some kind of difficulty, you know, when it comes to communication, when it comes to speech and language and writing and so on. And we need to do that automatically at scale, right? Because oftentimes, you know, guardians and teachers may not be paying attention, or it it somehow slips by, you know, where the child is growing up with some of these challenges and it has not been taken note of. So we need a screener. And then once you have identified the children who need some kind of help, what can artificial intelligence do to make the proper interventions? Yes. And that is the orchestrator. Right. So if I have identified a child with having a particular kind of speech difficulty, then can I now come up with AI tools which can help the child with in presence of a speech language pathologist, you know, in augmentation of what services a speech pathologist can do, you know, how can AI come in? So these are the two pieces, the screener and the orchestrator. And I I would like to add over here, Brian, that uh since you know we are talking about the post-Chat GPT era with generative AI, you know, our imagination itself has exploded as to what is possible, you know, in this new world of AI. And we have essentially looked at every aspect of that original proposal and seen how we can reorient now. And we have made several changes and we are making very good progress in both developing the screener and the orchestrator.

SPEAKER_01 11:03

Yeah. And for our listeners, Grand Island and some other districts around Western New York are partnering with the University of Buffalo to pilot and assist with their research. In particular, we're getting ready to pilot the screener. The screener is not evaluating, right? It's just looking for certain areas of need, and then it's a warm handoff to a human speech and language pathologist who will do the evaluation and consider whether or not that child needs additional work. But what if when that happens, and I think, ladies and gentlemen, as you listen to this, think about how in the old days or how we're used to now, it's an ad hoc approach, right? Like grandma might say to mom or dad, I don't understand your child, you know, my grandson and mom and dad say, Well, we understand. And then there's this conversation, and then maybe it leads to a pediatrician getting involved, and maybe it leads to evaluation. What UB has envisioned is a universal screening tool for every child, right, in the United States for free. And that is remarkable. And then the second phase is this orchestrator, not only assisting with uh the evidence-based research that's out there that might support Brian Graham as a student of the speech language therapist, but also maybe helping with the data collection and the progress reporting, which is such a tedious part of the speech language pathologist workflow, if we can relieve a little bit of that, then speech therapists can see more kids.

SPEAKER_00 12:36

That's such a good point you make, Brian, that it is also about efficiency. So if we can take out the grunt work that the pathologist has to spend in preparing data, preparing reports, finding out different uh tools and methods to address the students' needs. And different children have different needs. So there is a lot of time that goes into the preparation work. And if AI can come in there and make things more efficient, that frees up that much more time for the pathologist for a human-to-human interaction and personalization. So yes, so AI is essentially going to help in in all these three ways. Thank you for pointing out the middle third way of efficiency improvement.

SPEAKER_01 13:16

And when I when I talk about this, that's that's what hooks a lot of our professionals who are thinking about AI and how it can be embedded in their workflow. And what UB has envisioned, in my opinion, is exactly what is needed to support, right, not only a human-centered approach, but enrich the professional, enrich, and and by enriching the professional, enriching the student themselves, right? And the teachers. Yeah, this is this is such a great and important endeavor that you the University of Buffalo is leading the nation and again, the world in this work. And I can't uh thank you and your team enough for what you're doing.

SPEAKER_00 14:00

Yeah, and that cannot be overemphasized, actually, the fact that the human is always in the loop and it is a warm handoff. And uh, you know, it's essentially trying to make the job of the pathologists, of the human guardians and the teachers simpler. So that's where the idea is. And we are indeed uh leading the world. And here is an interesting connection since you brought up the story of postal automation in the beginning, is when we had envisioned this project, we were talking mostly about speech. Okay. Interesting. Yes, and then we brought in handwriting and that connects us to the postal automation.

SPEAKER_01 14:40

I I misunderstood you for a moment, but now I understand what you're saying. When you developed this grant for the National Science Foundation, it was really speech focused, but now you're looking at handwriting. Maybe, maybe tell our audience a little bit about that project because I think it's fascinating and it it does have that connection to your original roots of handwriting being identified at the post office. Now your project, you have a project where you're looking at children's handwriting, and and perhaps there may be elements of the handwriting that may lead us as educators down a path of considering is this child impacted by dysgraphia? Is this child impacted by dyslexia? Maybe explain that a little bit because that's brand new.

SPEAKER_00 15:23

Yes, that is brand new. And here is another interesting connection, and I'm sure you'll appreciate this that in today's AI world, you know, we talk about agentic AI and there is so much of machine learning that it's all about learning from examples and data. Okay. That's right. And then the old school, when computing was not so powerful, it had an element of human-driven features, which was not all just learnt from data, but there will be some rules that are crafted by the human experts. Okay. So in the postal automation days, go back more than three decades, you know, our algorithms would look at alphabet and we will look at few features and we would imagine how best a computer could interpret these, right? Right. And we have moved away from that particular methodology now where you just look at millions and millions of examples and the machine is able to learn and figure out the rules. But now with children's handwriting, what is happening, especially with this graph here, is every chill child could make a unique kind of a mistake in the spelling or in the way the shapes are made. And if you just go by data, there will be an overcorrection that will take place where you haven't identified the particular mistake the child is making. You are simply interpreting the word the child was intending to write. And so now we are actually going back to our roots and saying, look, this is how it used to be done, and a very interesting publication is forthcoming on this. And I like to say that, you know, as human beings, you and I and our teachers, you know, they are we are all able to use context in a very dynamic way. That means sometimes we need to use the context to interpret, sometimes we don't want to use the context. We want to actually see where is the child making the error, what kind of error. And this dynamic play is not that easy for the large language models driven AI methods. And so there is an interesting paper that's going to, you know, come forth on this topic. But uh we have been able to now go a full circle where we started with handwriting recognition and postal automation, and now we have come back to children's handwriting, yeah, and you know how what indicators are there for dyslexia and dysgraphia. And so it it has been a fascinating uh journey. Yeah, if you will.

SPEAKER_01 17:50

I absolutely concur. And it's just uh as as we're chatting, it's just another pathway that started just focusing on speech, right? And this other pathway has evolved that UB is helping to lead. And I imagine there are more. And one of the things I wanted our listeners to understand is how your leadership and the leadership of your team is intersecting with this Empire AI. And maybe you can give us an update on some of the funding that has been, you know, set aside for Empire AI, how UB is going to be leading the way, and how that project is going to support other SUNY and CUNY schools, I imagine, across the state.

SPEAKER_00 18:36

Yes. You know, for the listeners, you know, Empire AI is essentially a computing infrastructure which is available to a consortium of universities in New York State. Nice. And this is quite unique. Yes, it is super smart. I mean, kudos, uh very commend commendable for our governor to have this vision and for you know the various foundations who have all contributed, private ones, you know, to make this happen. It's a huge investment. It's about like$565 million investment. No other state can boast of this kind of investment in infrastructure. And also, Brian, you know, I've since we talked about what was uh prevalent three decades back in AI, today's AI, the foundations is all about the computing infrastructure. You need those powerful GPUs in order to do the cool things for public good. Right. And so as I had mentioned before, you know, we can reimagine all these problems, you know, through that lens. And what Empire AI is allowing us to do right now, and this is just as an example, is perhaps a few years back you wanted to automate, let's say you are looking at dyslexia or dysgraphia, or you're looking at speech, you would collect data from students, speech data, handwriting data, and you will have algorithms which will learn various aspects, and then you will have some kind of a decision-making, some kind of an automation. Right, right. Now, with Empire AI, the possibilities are endless. So we are not just looking at collecting the data, the pixels on a piece of paper, but we are looking at data of all kinds. You know, what is the lighting condition, what is the context, how is the classroom, who are his friends around him or her, and you know, what is the body language, what are the facial expressions, because all of those things weigh in on you know how a child is expressing or communicating. And uh so you can imagine the richness of the data and the tools all being possible because of Empire AI, because I was looking at some numbers. We have an alpha version of Empire AI right now on campus, which is about 104 GPUs H100s, I think, and then a beta which is going to be ready for use in a month, uh, which is about 228 uh GPUs uh uh Blackwell B200s, I think. So the speed up that you get from these is just phenomenal. Phenomenal. Tasks that used to take six months for learning are being completed in six days. Amazing. So that's what Empire AI is bringing for us, and and I think uh there is so much potential and promise, and hopefully, you know, in in the education space, we are going to have a huge impact.

SPEAKER_01 21:25

Yeah. So for our listeners, I've been blessed or fortunate to be part of this journey and very open to participating in research with the University of Buffalo. And our school system is set to pilot the screener. We're working uh with handwriting analysis, just as uh Venny was talking about. But maybe if from from the university's perspective, uh speaking to educators listening to this podcast today, maybe you could help uh assure other school districts that they should consider. partnering with UB for this research because the amount of work that goes on behind the scenes to get this approved is incredible. So maybe just share a little bit.

SPEAKER_00 22:08

Absolutely. And we are very mindful of the fact that in general, you know, we must make sure that privacy is maintained, uh data is anonymized, and there are IRB protocols, you know, which we go through to make sure that the data is collected, you know, in a proper manner. And so we want to assure uh schools like yourself in Grand Island and we also have partners at Sweet Home School District where we are trying to collect data. And we are being very deliberate in how we proceed on this. Because as you know Brian, we'll probably get a one chance to make this work with the schools because once it doesn't work well, you know, people are turned off and then it's difficult to gain that trust back. So we are going trading this very carefully. Hopefully within the year, you know, we will have some pilots in the schools and success. But we absolutely want uh more schools to come forward. We have some partner schools with our other university institute members in Reno Nevada and in Seattle Washington. We are looking for more schools to participate. The more the data the better it is the better these algorithms. So you know absolutely and and that will be I think the the final what should I say evaluation on you know how successful this this whole project has been is when we can put these tools in the hands of school teachers and they see a benefit you know from these.

SPEAKER_01 23:37

Yeah oh my gosh so for listeners uh feel free obviously you can reach out to me I can put you in contact with Joelle or other people in the university to learn more to see how how this research can benefit your children in in your schools it's truly for the social good and there's nothing better than that. So now that we have entered maybe this generative artificial intelligence phase this this trajectory from machine learning to where we are now and what's what's over the horizon in the future maybe uh share with us how this work I mean your the tit your title is Senior Vice President of Research, innovation and economic development this trajectory of how AI is being used for the social good in the leadership of UB is also part of the economic workforce of our region. What are some some things that you see in the future that get us excited get our students excited about this work?

SPEAKER_00 24:42

Yeah in my mind you know this is a great differentiator. University at Buffalo stands out as one of the few places I would argue anywhere in the world where you can come to a university campus and have access to this kind of computing infrastructure. That's right. And this is being supported by by our state and we are looking at primarily those projects which are for public good. You know and that also gives you a different level of satisfaction when you go through these projects. So we are looking at a whole bunch of projects in the health space of course in the education space in the environment space and so on. And this is essentially going to affect the ecosystem the workforce development the economic development of the region I know for a fact when I go internationally and I I just didn't did that a few weeks ago. You just came back from India. Just came back from India and when you talk about you know to talk to students you know who are considering various university offers you know this is so attractive to them. This is such a big draw because AI is happening they can all see it they can see it in their phones they can see it in their daily lives and to be able to be part of some very impactful projects and gain hands-on experience you know UB does provide them with that opportunity and then also you know the ecosystem itself there is going to be a cascading effect because you know the students and faculty are working with companies many of our local companies they can work through us you know in partnership and frame projects you know that can be run on Empire AI and you know come up with some solutions that are quite interesting. I do know that there are faculty members who are also attracted to UB because of this differentiator we have and you know I would say that in in the next couple of years uh we will definitely see a very positive impact of of this endeavor.

SPEAKER_01 26:38

Yeah I've also noticed that many of the undergrad and graduate programs that are offered through the university have embedded artificial intelligence into that workflow, into that learning flow. Maybe just share from your perspective how how a few years ago when I came to that retreat that was not the case. And now three short years later or you know four ish we're seeing the university embrace this in many different departments.

SPEAKER_00 27:07

Yes because uh AI is touching every discipline there is and at UB what we have done is we have paired the AI technology with the ethical social responsible behavior that is required in every discipline and we have launched a new department you must have seen the press release on that. That's right it is called AI and society. That's right and very interestingly it comes up with new degrees in the format of AI plus X. Yeah where X can be geography, X can be linguistics, X can be economics, X can be philosophy. And I also want to say that while we call it AI plus X, it is the X which is more central in that whole degree program because it is the discipline that is going to drive that ethical and responsible framework. So AI is is much more than just a STEM field, much more than just computer science. It is for every discipline you know in in higher education I think this is what the trend is going to be. And once again you know I'm very proud to say that UB is leading the way we are one of the first to offer you know AI plus X degrees with a new department.

SPEAKER_01 28:22

Our Dr. Michael Capuana he is the leader of Erie One Bose's and he put together a tour for all the superintendents and we just did that a few weeks ago and we came here to the campus we started off in the business department and then we came of course to the AI Institute for exceptional ed and it was remarkable I think a lot of my colleagues were so impressed and it was really an eye-opener right for us as educators to understand maybe for us as educators to start thinking about what what would does the next 13 year pathway look like for a brand new kindergartner coming to us in September and then 13 years later you know how what should we as educators be doing at the public you know public school private school side to get that kindergartner ready for a new frontier a new a new economic workflow 13 years, 14 15 years from now, right?

SPEAKER_00 29:18

So yes I I mean in this space I think many of us have realized that making any prediction is not going to age very well. No five years back none of us knew you know what post chat GPT would look like. In fact if you look at the interviews of some of these big tech giants including Anthropic and Google and Microsoft and OpenAI, many of them themselves did not predict that is true you know the way AI has taken off. So I think it's going to be exciting as long as we have the guardrails to make sure that it is going to be done for public good and we are vigilant to make sure that the abuses you know are curtailed right at the bud, you know, I think we'll be okay.

SPEAKER_01 30:00

Aaron Ross Powell Yeah I do too and I think to that end as educators of our of the youngest children coming to school I think that has to be the centerpiece, right? That has to be what we focus on ethical, responsible trustful use in AI throughout the the journey of school and then beyond. So I think there are a lot of lessons to be learned and it's so great that UB is leading the way well I am always thrilled to be here at the University of Buffalo and to have you on the podcast venue as we wind things down. What are some final thoughts you'd like to share with our audience?

SPEAKER_00 30:38

Well it's always great to be part of this conversation Brian and I look forward to our meeting this has been a great tradition.

SPEAKER_01 30:44

Yeah so let's do it again in a year. I love it. I love it so much and uh for our friends listening today don't forget that if you are interested in partnering with this important research I'd like you to reach out to Joelle and Joelle's email address is JES46 at buffalo dotedu. So that's JES46 at buffaloed edu and she can then connect you to the appropriate department where you might want to explore how to be engaged and pilot and be part of this amazing journey that UB is leading the world in. So thank you so much and don't forget ladies and gentlemen if you are if you like achieving joy and mastery in public schools you can follow or subscribe on your favorite podcast platforms such as Apple, Spotify, Odyssey and so many more until next time thank you so much for tuning in thank you for joining us today as we explore the concept of achieving joy and mastery in public schools. We will be interviewing school districts in Erie and Niagara counties on a regular basis as we look to shine a spotlight on all of the amazing programs and practices essential for achieving joy and mastery with our students, faculty, staff and community this podcast is sponsored by the Erie Niagara School Superintendents Association and we hope you consider subscribing