Review the latest events anytime at cspan. Org coronavirus. The u. S. Chief Technology Officer talked about Artificial Intelligence and competition from china. Hosted by the hudson institute, this is about 50 minutes. Okay. Well, welcome, everyone, to our meeting on ai and advances in Quantum Information Sciences or quantum computing, et cetera. Im pleased to have an opportunity to welcome michael, who has a thoroughly intimidating title of the chief Technology Officer of the United States. So its i think weve come to the right place to engage this this important subject. As as you know, the the administration has taken a very energetic position in trying to advance the stateoftheart and above the applications of Artificial Intelligence and the Quantum Information Sciences to our nondefense applications. Theres there is parallel efforts of substantial scale in the department of defense, through the Defense Advanced Research projects agency and other defense entities because these technologies are, indeed, of universal applications. So i like to take an opportunity to begin the discussion with michael to, perhaps, start with a bit of current news. Where europe has european union, in particular, has tabled some of their ideas about how to manage some of the particularly, the ethical issues relating to the use of Artificial Intelligence. Weve had, of course, quite a number of years of tension on a transatlantic basis about the application of advanced technologies to commercial, as well as governmental, application. So be interested in getting your take on this, michael. Thank you. Yeah, absolutely. Well, thank you so much to hudson for having me. And thank you for this d this conversation. Yeah. So yesterday was was was a big day of news for the world of ai regulation. As you probably know, United States was kind of first out of the gate in january where we proposed the airegulatory principles and those are out for comment. And then since then, the eu, yesterday, has released their attempt to do what we did in january with providing some sort of structure around the way they are trying about regulations of aipowered technologies in the european union. So i think our kind of take is sort of twofold. I think, one, you know, were very encouraged to see a lot of focus in their document on the importance of fostering an Innovation Ecosystem that is friendly to Artificial Intelligence technologies. They talk about the importance of developing and working on research and developmenttype projects and helping drive startup squs sma startups or smes as they call it. And they also talk, very much, about a valuesbased approach. One that the United States expressed and we put out our principles in january. But the one thing i do think is important to flag and and and one thing where i think i think there could be some room for improvement and implementing it. What we found what they actually put out yesterday, really does in some ways sort of clumsily attempts to bucket aipowered technologies as either high risk or not high risk. So the proprososed way their sce is structured, there will be some sort of group in europe of some kind that will make some decision on whether a technology is high risk or not. If you are high risk, you have to go through a pretty extensive regulatory approach. If you are not high risk, you dont have to do anything. We believe this sort of allornothing approach is not necessarily the best way to approach regulating ai technologies. We think that ai regulation is best sort of serves on a spectrum of sorts. There is certain types of technologies, aipowered technologies, which will require heavy regulatory scrutiny and we, in the United States, are prepared to do that. But there are quite a few that need just a little or not at all. And i think creating the spectrum is important. So i think thats kind of where our biggest concern is. And ill be traveling to brussels next next march, next month, and speaking to folks over there and kind of sharing some of these concerns. I think there is a lot we have in common. But i think this approach of very sort of bluntly bifurcating the entire the entire ai tech ecosystem into two buckets is a little bit harsh. I think its a very good point. And as may know, i was also served on the dod defense science board. And we have done studies on the application of modern technology to the development of both ai and quantum. And one of the opportunities that may evolve as the technology improves, to facilitate better linement wialh our allies in europe, is an effort thats underway with in darpa to develop explainable ais. So the user of the outcome or output of an aibased bit of analysis is able to understand the coupling between the outcome and the the data that produced the outcome. And were, of course, not there yet. But that there may be some opportunities for Research Collaboration between the u. S. And and the eu to, perhaps, better solve this problem of of explainable ai. Yeah. I couldnt agree more. I think it sort of its sort of a its a good segue into sort of how the u. S. Approach, in some ways, differs a little bit to the to the european. I think when we put out our principles last january, they sort of focus on sort of three main themes. And i think one of them hits directly on this explainability question. On the first go around, i think the most important thing in the u. S. Model is public engagement. Whenever we attempt to pursue any kind of regulatory action of any kind, you know, we, as the, you know, federal government, have some experts. But the community is the people who know this best. They have scientists and technologists and experts who can help us think through that. So there is a lot of emphasis in our framework on public engagement. The second key is really around limited regulatory overreach. We believe we need to create a model that is risk based and use based and sector specific. So the types of regulations that you may have for an Autonomous Vehicle or for a drone is very different from the type you will have for a medical diagnostic. And rather than bucketing technologies, there has to be a bit of a spectrum and flexibility in the model so that you are able to actually regulate appropriately for the risks each of those technologies provide. And the third, which i think you bring up very astutely, is this idea of promoting trustworthy ai. This is something we deeply care about. And we need to engender trust in the American People and the technologies that they are using. And we need to create a regulatory model that allows that trust to be built. And having better r and d and Better Technology around issues like explainability, i think, will get us a lot closer to that place. One of the issues thats related to this thats coming fast upon us is the interaction between ai and the internet of things. Science board has been doing some work on the technologies of autonomy and counter autonomy. And obviously, Ai Technology is one of the things thats going to make iot work for the whole society. So there may also be some opportunities for collaboration with our eu colleagues on trying to understand how we will manage the introduction of iot. Because, like other applications where ai is involved, the range of applications is extraordinarily diverse. I couldnt agree more. And i think where that you know, where we have sort of manifested that type of thinking in our approach is actually through the first large white house summit we held on Artificial Intelligence was titled ai for american industry. And we were trying to express when we were hosting this event is that Artificial Intelligence is going to touch every industry in the United States. Whether you are doing Oil Extraction in texas, whether you are doing farming in iowa, whether you are doing biotech in boston, you will be using this technology to drive your business. And if the u. S. Wants to lead the world and win Artificial Intelligence, we need to make sure that all of those industries able to capture the benefits that Artificial Intelligence can provide. So being able to have that kind of very important dialogue with our allies around the world with how we can move all these Industries Forward is absolutely critical. Right. One of the, lets say, news items thats been a pretty constant drumbeat for the past half dozen years or so has been chinas alignment of its perception of its national and security interests with investments in advanced technology. They announced their made in china 25 initiative five years or so ago. Which identified about ten areas of technology that would be getting particularly high investment from china. And its often described in the terms of billions of dollars. And id be interested in your observations, michael, about how chinas efforts have been coupled to the administrations initiative and your perception of the chinese effort. Yeah. I think theres sort of two two threads there to pull on. I think the first, which i think, you know, needs to be said and i think, to me, is should be evidently apparent and probably communicated more often generally. Is that, you know, if a chinese companys party is using Artificial Intelligence to track people in their country, to imprison ethnic minorities, to push forward a complete surveillance state, to maintain a great chinese firewall and essentially restricting the content that Chinese People have access to, these are the use cases of Artificial Intelligence that are are deeply in conflict with western values. And this is something that we have tried to communicate, and continue to communicate, with our friends and allies in europe. And there has never been more of an imperative than now to ensure that the u. S. And our allies lead the world in Artificial Intelligence. We need to ensure that the next great Technological Breakthroughs are made here, in the west, and are underpinned by western values. And i think if we dont lead, we run the risk of these values that are diametrically opposed to everything we believe in, slowly permeating these new technologies and then being export abroad. So that is why the imperative is so great. That is why the president signed into signed an executive order launching the american i initiative. That is why we have pursued a whole of government approach the last three years. And that is why we made the big announcement last monday that many of you saw that we are committed to doubling nondefense ai spend by the federal government in the next two years. That is moving from about a billion dollars to 2 billion of federallyfunded r and d in ai in the course of only two years. This is a massive and incredible step forward in our commitment to American Leadership in this in this particular domain. And i think whats very important to remember and so thats point one. I think, point two, what you brought up, is sort of these commitments that have been sort of publicly asserted by the Chinese Government. You know, we believe that, you know, theres a lot of we have a lot of skepticism in the validity, voracity of those particular statistics. And i think i challenge we have a lot of brilliant thinktank people here. Lots of journalists here. I spend all to spend more time thinking about if you are attempting to report on an action taken by the Chinese Government to spend more money in ai, is that actually happening . Like, really, are they actually spending billions of dollars . Is that really a true statement . Can you compare that number to the number that Congress Appropriates and has actually spent and put out the door by our agencies . I think the short answer is no. And i think there needs to be a more and stronger and better and bigger conversation around validity of those numbers. I point you to two studies that came out. Georgetown. A great institution. A team put out a study last december that cast doubt on a lot of these numbers. And said that, in reality, theyre definitely not spending tens of billions and spending a lot less than that. And thats the type of narrative that we need to make clear. Because when we are trying to make comparisons around what the west is spending and how we are approaching our r and d ecosystem, we actually need to be comparing apples to apples. Its a good point. And one of the things i think, ironically, is going to render chinas investment less successful is that, that parallel with the their made in china 2025 initiative, they also have emphasized what they call Civil Military fusion, which is an effort to extract the military applications of these advanced technologies. Both the Quantum Sciences and ai are technologies that will have universal applicability and trying to force feed the scientific effort into producing military advantage will have the more likely outcome that will produce neither military advantage, nor advance the underlying science. So it it is something of limitation. And so i think were were likely to be more successful with this approach. And i was very reassured by your observations about the scale of the increase. And having previously served in the office of management and budget as an official there a number of years ago, one of the questions i always ask about Public Sector investment, which is easy to measure, is what are your expectations for the outcomes of this investment . And are there pertinent metrics . Are there formed expectations that might help shape public expectations about the scale of the investment . Absolutely. And i think the way the best way to answer that is to kind of give a little bit of a description of the type of Innovation Ecosystem we have here in the United States and the role that the federal government and its agencies plays in driving innovation broad broad broadly in the United States. So whats very different than almost any country in the world and particularly in china is the way the federal government spends research and development dollars. We dont have a ministry of science that has dolled out x number of dollars. We have research happening across all our federal agencies. Darpa, for example, is doing incredible work. You have National Science foundation which is appropriated roughly 8 billion a year to invest in earlystage, basic Research Done by at at a lot of our universities. There is department of energy that has billions of dollars that are spent through there. Incredible national Lab Infrastructure with National Institutes of health which does a lot of our biohealth related research as well. Each have their sort of own aspirational goals and pieces of the puzzle they play. In some ways, were incentivizing, creating free market of ideas around innovation. And the part that the federal government plays in the larger spectrum is that generally speaking, the federal government is investing in early stage, precompetitive, basic research and development. And thats very different than what the private sector does. And thats thats by design. So the the types of research that the federal government, generally, approaches is the type of research that the private sector is not incentivized to do on their own. That is a gap we try to fill and do it in a way where these ideas can come to life and taken to the sector and brought to fruition and ultimately to commercialization. And i think a great example to show how our system is unique is this breakthrough that happened last year on quantum supremacy. So there is no doubt in my mind that someone in beijing had sort of, you know, called upon someone someone else in china to achieve quantum supremacy before before United States did. You know, we didnt make that call here in washington to our kmunl community. Yet, the United States made that breakthrough first. The federal government in this early stage basic research. We made a commitment years ago to investing in a quantum lab at uc santa barbara. Its doing incredible work and we continue to fund their ear early, basicstage research. Some breakthroughs were made. Google saw this. Said, wow, this is a great team. We could bring them on board. We could equip them with more resources we have. More compute time. Whatever. They acquired essentially the group. Brought them in house. And that group was able to achieve what they believed was quantum supremacy. Now, they had to prove that their device could actually be faster than a traditional computer. So who has the Fastest Group of computers in the world . The federal government does at our national lab. So then they took their, you know, their breakthrough and they went to the fastest computer in the world thats run by the d. O. E. And they ran the test to prove they had actually done it. So here you can see this incredible sort of Virtuous Cycle of all pieces of the ecosystem working together to federal government doing basic funding work at academic institutions, moving into the private sector, then having to go back to the federal government ult mayimately