The National Science Foundation wide-and-far on artificial intelligence

The National Science Foundation (NSF) is on the march, establishing what it calls regional innovation engines. It is getting organized around AI.

The National Science Foundation (NSF) is on the march, establishing what it calls regional innovation engines. It is getting organized around artificial intelligence. It’s getting and granting-out more money than ever. For an update, the Federal Drive with Tom Temin spoke with NSF director Dr. Sethuraman Panchanathan.

Interview Transcript: 

Tom Temin All right. I want to start with the NAIRR. This is the National Artificial Intelligence Research Resource, a concept for a national infrastructure that connects U.S. researchers, I’m reading, to computational data, software model and training resources. Tell us what’s actually going on with this.

Sethuraman Panchanathan As you all agree, AI is part of everything that we are doing today and it’s probably going to occupy more of a central core. It’s almost becoming like electricity is today where you don’t even pay attention to it. It is part of your regular daily life. So you can think of AI in that terms as you think of the future. So the NAIRR, the National AI Research Resource, was essentially put together in order that we can move fast and build things for our nation. So that’s a very important thing that at this moment that we need to invest in, and we need to make sure that we are unleashing all possible ideas, talent, innovations everywhere across our country. This research resource will serve as a critical driver for a healthy, trustworthy U.S. AI ecosystem. And as you said in your remarks, it will strengthen the pursuit of new AI tools, systems, and fields of studies. Whether the researchers are in academia, local or federal government, or small businesses from every corner of our nation. Now, this will be critical for our nation and its continued leadership at the edge of AI innovation.

Tom Temin There’s a lot of resources that are named here. Computational data, software. Where will all this come from? What are the sources of the raw materials for people to conduct research with?

Sethuraman Panchanathan Now, as you can imagine, that so far, if you look at the AI development and as you think into the future and the amount of resources and all the other things that comes with it, not only the compute resources, the software resources, the model resources. And all of those will require investments if you want to democratize this access and therefore be able to provide the ideas and innovations that’s everywhere across our nation unleashed. So the pilot, as a first step is a proof of concept for the NAIRR. And how do we take the the steps in terms of realizing the full vision of what it takes. And what that would entail, obviously, a several component parts that will make up what a NAIRR would look like. We need high performance computing resources, we need commercial cloud computing resources we need data sets, access to data sets covering a range of domains that can support the development and testing of AI powered systems. We need, of course, advanced tools, software and platforms. We clearly need AI models that have access to open and proprietary AI models so that we can facilitate research on the models themselves and applying models to novel problems. Of course, most importantly, we also need learning and training opportunities, the education opportunities that will enhance researchers skills. So all of these components have to be advanced through NAIRR, and the NAIRR pilot therefore, in partnership with over a dozen agencies, federal agencies and over 25 nongovernmental partners, whether it is industry, whether it is philanthropy, whether it is nonprofits, all of them are coming together so that we might build this pilot. And therefore, that forms the basis for the full scale deployment of NAIRR. And so that’s what we are looking at. So the pilot implementation essentially will bring together computational data, software, model, funding and training resources from a range of partners within government, the technology sector, nonprofits, philanthropies demonstrating the shared priority of addressing the critical resource gaps in the AI community.

Tom Temin And what is the mechanism by which this group you mentioned? There’s almost 37 parties to the pilot. Who decides where the supercomputing comes from, where the models come from, where the data training comes from, and this kind of thing. Operationally, how does it work in a practical sense?

Sethuraman Panchanathan So NSF has been assigned as the the management agency, which brings together these partnerships. NSF has got a track record, as you know, of bringing together partners of various types, interagency partners, industry partners, partners from the nonprofit philanthropy and etc.. We have got a proven track record of how do we bring everybody to the table, co-create and implement together for the benefit of humanity, society, and in this case, advancing the research ideas and building the talent and building the innovative outcomes that we all seek.

Tom Temin We are speaking with Dr. Sethuraman Panchanathan, director of the National Science Foundation. Well, if say, the Energy Department says that’s the one where the supercomputing will come in, just to make an example. Who pays for it? And is there money available for that extra computing that they will be obligated to provide to the NAIRR?

Sethuraman Panchanathan So right now, what we are doing is we are all co investing with the resources that we have. And you point out the Department of Energy, a great partner bringing together computational resources. But we are also looking to this pilot expanding to a full scale for which additional investments will be needed for all of us to be able to make sure that we are scaling all of these resources to meet the demand, to meet this moment that we need to meet. And so and I’m delighted that with leadership of the administration, the Biden-Harris administration, and the true bipartisan support that we have in Congress, we have both on the Senate side and the House side, Senate AI committees, they have had conversations with a variety of folks, their panels and so on. And so through that, the create AI act, for example, is a way in which we are looking at how we might invest and we are looking forward to the congressional investments, because this is a very important moment, important time, and we need to be in the vanguard of these ideas, talent and innovation and then keeping us ahead of our competitors. And therefore, we are working diligently, as a collective force, bringing together all of the folks that have a stake in this. And I am pleased to say that I’m excited by how things are progressing, and I look forward to a wonderful outcome from all of this.

Tom Temin Yeah, it’s been up and running only about a month at this point. The pilot. Anything has happened yet?

Sethuraman Panchanathan Tom, I’m glad you asked this question. On Oct. 30, 2023, President Biden signed the executive order 14110. And the agencies were all tasked with various things. And among the tasking for NSF, there was a directing to NSF to launch the NAIRR pilot within 90 days. And I’m proud to say that within such a short time, we were able to bring all of the parties together, and that is a credit to all of the participants were all excited and wanting to work together. And so we have already launched the pilot. And the pilot is, as you said, the first step in seeing how we can advance on all of the areas that we talked about. But we are going to have more convenings, we’re going to have more releases of opportunities for people to participate through the resources that have been identified so that people can then bring their ideas and make sure that the resource gap that I talked about, the critical resource gaps in the AI community are addressed, and therefore we can have participation from a broad community, and that will help us build these new models, new algorithms, new tools and new innovative outcomes.

Tom Temin So your sense is that there are a lot of ideas out there about AI that could be developed by people, if only they had resources they can’t access now. That is to say, if you need substantial cloud computing resources or you need substantial supercomputing resources, now there’s a pathway to those things for those, let’s say, underserved researchers.

Sethuraman Panchanathan Absolutely. Underserved locations, underserved researchers. This is about, as I said, democratizing AI access, AI development. AI idea is being unleashed at speed, at scale all across our nation. There’s so much that we can do here. And at NSF we are also focused on development of talent as you know. We are the STEM talent agency and we invest quite extensively K-12 investments that we make, community college investments that we make in skills development, as well as investment to be making higher institutions and continuous upskilling, reskilling kind of investments. These are all ensuring we have a program called expand AI, for example, through which we are investing, particularly in institutions that have not had a chance to participate yet, like the minority serving institutions. And the second tier of research institutions, as well as the first tier of research institutions, all of them being enabled. Minorities having institutions, tribal colleges. K-12, as well as research universities, all of them participating because we need all of that and more in order for us to be able to deliver the kind of the scale of talent that will be required for us to be in the vanguard of competitiveness.

Tom Temin So it sounds like one of the first orders of business for this NAIRR consortium, if you will, is to develop the criteria by which you will grant access and resources to those that apply.

Sethuraman Panchanathan Correct. In any of these, you want to have a set of criteria by which, when people are asking for resources  for the development of their ideas or building their ideas or building tools and so on. You want to make sure and as you know, again, NSF and the other agencies are very, very good at that in terms of looking at those ideas. Looking at the strength of the ideas and the potential impact that those ideas can generate. And therefore we use the criteria by which we choose those projects that we want to provide access to these resources so that they can advance very rapidly, in terms of the development of those.

Tom Temin Just give us a sense of who these 25 non-governmental partners are. Are they mostly universities or who else might be part of that?

Sethuraman Panchanathan No, these are industry partners. Let me give you an note. This is like if you take the who’s who in the AI industry, Nvidia or Microsoft or Amazon or IBM. Open AI, the who’s who, their happy to share the list with you, but the who’s who are all part of this and because people want to be part of this. They all want to make sure that the talent and ideas that are out there are inspired, motivated, nurtured, and brought to life so that they can still advance on their industry priorities and what they want to do in terms of all of the stuff that they are working on in AI, whether it is the development of the next generation of hardware for implementing things, anthropic, AMD, Databricks, l’autre AI, Google, Hewlett Packard Enterprise, Huggingface Intel, meta. I can go on, Omidyar Network.

Tom Temin I think we get the idea.

Sethuraman Panchanathan The point that needs to be made as the who’s who in AI are all participating in this, and that’s the exciting part. Because they all believe that this is important. The democratization of AI resource is important. The inspiration of talent across our nation is important, and because it also ensure the success of those entities, and that’s what we are trying to do. It’s a whole of nation approach. The whole of government expanded the whole of nation approach, if you want to look at it that way.

Tom Temin We are speaking with Dr. Sethuraman Panchanathan, director of the National Science Foundation. And I want to switch gears here and talk about the regional engines, the innovation engines, I guess there’s ten of those. They were awarded last May and now they exist. There’s a map of them and so on. These were funded by Chips act money primarily. Tell us what has happened since then and what do you expect out of these engines. And we can get a little bit into the anatomy of them also.

Sethuraman Panchanathan Thank you so much, Tom. The last time that we spoke, when it came to NSF, I really was convinced that the amount of innovative potential that is there across our nation everywhere should be unleashed. And so we launched this new Technology Innovation Partnerships directorate, the first new director of NSF in 31 years. And as soon as we launched the directorate, we have a number of exemplar programs on how innovation can be made possible anywhere across our nation. One exemplar program is the regional Innovation engines. How do we take the innovation in place that is available all across our nation? How do we bring together a mindset of multiple parties working together so that they might build this innovation ecosystem through public private partnerships? I always liken this to the lab like thinking. How do you have academic researchers, industry practitioners, all of them working together in a way that you unleash not only fundamental ideas and technologies, but you also build them and make possible innovative outcomes, industries of the future. And also, more importantly, the people will become leaders of today and tomorrow.

Sethuraman Panchanathan So I’m excited to say that the reason innovation is in competition at two scales. One is the type one, which is about seeding innovation. And we were thrilled because when we put this call out as a concept note that we wanted people to present in terms of what their region is best at, in terms of exemplifying that innovative in place, potential that was there. And we got 670 concept papers which spanned all across our nation, all territories, all regions, all states participated in that. We made that publicly available people to look at it. And through that, people, again formed strong partnerships. We had a couple hundred proposals. We funded 44 of them in the first tranche right away, each $1 million. But every one of those regional innovation engines, I’m excited to say, is unleashing the innovative potential in various parts of our nation. What is it states in this case. And then we had the type two, which is a little bit more mature innovation, ecosystems being invested into scale and rapidly have the innovative outcomes. And again, through partnerships. And we just launched the ten innovation engines all across our nation. We had a tremendous repsonse again, we have 36 semifinalists, 16 finalists, and we launched ten of them. And you look again at all of these, you will see that this is not just, we are hoping that something will happen. These are innovation ecosystems that are ready to unleash the ideas, talent and of course the innovative outcomes through industry scaling, new companies, forming, entrepreneurial ventures being launched. So it’s very exciting to see this. And, again, there are many, many examples of that I was in. Winston-Salem, for example, with the first lady, Dr. Biden talking about one of those, which is on regenerative medicine, like it was artificial kidneys and the companies and the ecosystem of H2 partners coming together. All the community colleges, universities, the research centers, the industry, the the governor’s office, the economic government ecosystems. It’s just amazing to watch that.

Tom Temin In looking at the map of the regional innovation engines, there are not California, there’s not Massachusetts Boston area, and there’s not New York City. They are sort of a midwest, almost southeast orientation. So it sounds like you’re smearing out the expertise and the resources somewhat so that it doesn’t have that Silicon Valley flavor, which is getting a little bit green around the edges.

Sethuraman Panchanathan We all know Silicon Valley, Kendall Square and many parts of the nation are vibrant, robust. We want them to see how they can further our investments. We have these investments in these universities, in those regions, startups in those regions and so on. And they will continue to be vibrant and do amazing things for the nation. But I always say that unless innovation anywhere vision is realized, unless opportunities everywhere vision is realized. And this is what I keep talking about since coming to getting NSF, innovation anywhere, opportunities everywhere. We will not be able to outcompete all the things that we need to be doing in this hyper competitive environment. But even more importantly for me is that we have the latent talent everywhere across our nation. How can we not have all of the talent in our nation inspired, motivated, unleashed and make prosperity possible for all our citizens everywhere across our nation. And I’m convinced, and this innovation engine is a proof point, that yes, it is possible because there is that excitement everywhere. There is a commitment everywhere. There is this mindset of people coming together as an ecosystem and wanting to unleash talent, ideas and innovation. As I keep saying this. And I’m very heartened by the fact that this is not just a hypothetical thing. This is reality on the ground, making amazing things possible. And I’m confident that with the partnership with the Department of Commerce and the regional technology hubs and with the partnership that we have with venture capital firms, the partnership that we have with the economic development ecosystems across the country, that these are not just only what NSF investments do. Quite the contrary. We invested $150 million in these ten for the first two years. I full-scale it will be 1.6 billion. But what I’m excited to see is that 115 million is matched by $350 million of co investments made by these partners. To me, that shows commitment. To me it shows the trust that this is going to work, and to me most importantly shows that we are going to have a sustainable, scalable ecosystems develop for the future. And that’s what makes me feel very, very excited. And I’m very grateful to be partners and very grateful for the participation. And at this moment, as we keep talking about it’s an important moment and I’m excited by the bipartisan support. I look forward to seeing how the authorizations of the science portion of chips and science is made into appropriations, so that we can meet this moment with the investments that are needed to unleash these possibilities.

Tom Temin My guest is Dr. Sethuraman Panchanathan, director of the National Science Foundation. And you have put quite a lot of resources, let’s say, money people, talent organizations into these salads that are these regional centers. What will be some tangible outputs that you’re hoping for?

Sethuraman Panchanathan Now, there are several metrics by which we look at this. Clearly jobs that will be created, the new industries of the future that will be created, new ideas that will be birthed. And there were new companies that will work, and existing companies being able to take advantage of those ideas and scale their operations and position themselves for the future. So it is about the industries of the future. The entrepreneur ventures, the future. It is about the most important element, which is talent for those industries to be successful in the future. And then it is about, of course, the talent then becomes the manifestation of that is the jobs of the future that is made possible through this. And it is about the vibrant environment everywhere across our nation. So people feel that wherever they are, that they have the opportunities, that they can exercise that talent to the fullest extent. So all of these, right, the hard metrics that we have that are also, in tangible soft metrics also. Sometimes all of these are important to keep in mind as we are thinking about the advancement into the future and measuring our progress.

Tom Temin And do the regional engines actually have a physical locus? Is there a frosted door somewhere that says Regional innovation engine here?

Sethuraman Panchanathan Yeah. So I think the coming together. If you look at these regional innovation engine locations, you will find that there are these if you want to look at them as a nodal presence. But it is not just all centered only in the North. It is obviously it will be successful only when the nodes are connected to all the spokes, like the spokes in a wheel. The spokes, all of the other nodes that are in the ecosystem. So they are all working together as a network rather than just one central node.

Tom Temin And then there are quite a bit of sub locally let’s say, around that are subordinate to those engines. What happens at all of those?

Sethuraman Panchanathan No, I wouldn’t call them subordinates. I would call them as partners. All of them bring together unique perspectives that enriches this overall concept. And so, for example, if it is a textile ecosystem, that might be one part of it which might be focused on the next generation of smart fabrics. That might be focused on dyes and chemicals that are necessary for the textile ecosystem to be successful. So each partner brings in unique skillsets. And what is exciting is that the community college and the university ecosystems are saying, what is the skilled technical workforce that is needed for these universities, for these companies to be successful? What kind of mindsets do we need to have the entrepreneurs who would then be birthing new companies so that they are participating in that form, then it really guarantees the success.

Tom Temin And you have finally described a pair of really ambitious programs, richly funded, but also partnership money. Do you feel the NSF has the institutional capacity to make sure that you stay on top of all this? Because it’s really a major expansion of activities.

Sethuraman Panchanathan I’m glad you mentioned that Tom, because if you look at the birthing, we got $1 billion of additional resources. It was thanks to Congress as appropriations, FY 2023 appropriations, we got $1 billion for the new directorate of Technology Innovation Partnerships, close to $1 billion. So that is the kind of investments that we need to make in order to build the capacity that you’re talking about to be all across the agency. It’s not just only in the directorate. This billion dollars is to build capacity, because at the end of the day, it is the success of the directorate and these programs are only when they leverage all the amazing innovations that’s happening to the investment that we are making in all the directorates, and then that this work that, again, fuels more progress and discoveries in the directorate. So it’s a complete and again, to use the word ecosystem. It is an NSF ecosystem if you want to look at it that way, that is responding to this in a holistic manner. And therefore this investment of $1 billion therefore helps us move that needle in a capacity that you talk about the capacity way that we need to be responsive. But the scale, as I said, the appropriations need to follow through with the authorizations that we have from the science bill, chips and science bill. The science part of the authorizations becoming appropriations is only allowed to scale and make this really possible with the amazing interest that we have seen all across the nation.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.

Related Stories