Minute, as a matter of fact. Good afternoon and welcome to the National Press club. The place where news happens. Im mark, chair of the board of governors, and Senior Reporter for investment news. Thank you all for joining us, both here in person and online, for our headliner event with s. E. C. Chair gary gensler. We are happy to accept your questions and after chair genslers opening remarks, ill ask as many as time permits. Mark to submit a question online, please email headliners press. Org, and put gensler in the subject line. For our cspan and public radio audiences, please be aware that in the audience today are members of the general public. So so any applause you hear or reaction is not necessarily from working press. It is now my pleasure to introduce our distinguished head table. Please stand when your name is called. Starting on monday right at the far end, mark, senior editor, cnet money. Next, j. D. , a reporter and host at the street. Com. Next to j. D. , con stance of con stance marks productions. Next we have miriam, technology reporter, american banker. Next to miriam, kimberly, host and senior washington correspondent at marketplace and my immediate predecessor as chair of the National Press club board. Next to kimberly, mark, a former Club President and senior economic analyst and Washington Bureau chief at bank rate. Skipping over todays speaker for a moment, but well definitely get back to you, chair gensler, lori, president of Stanton Communications and our Headliners Team leader. And one of our Headliners Team leaders. Now to my left, eileen, National PressClub President and styles and standards editor at axios. Were having a terrific year so far under eileens leadership and the turnout today is a symbol of how were coming back strong with inperson events during eileens year. Next to eileen, emily, National PressClub Vice President and washington correspondent at cnbc. Next to emily is scott, correcter of the office of Public Affairs as well as counselor to chair gensler and a guest of the chairs. Next to scott is lydia, s. E. C. And financial regulations reporter at bloomberg. Next to lydia is andrea, white house and economics correspondent at reuters. Then we have emily, senior Vice President , American Investments council. And at the end we have stephanie, director of speech writing in chair genslers office and a guest of the chair. [applause] mark this is not my first rodeo, s. E. C. Chair gary kpwepbsler told reporters in an appearance before the Senate BankingCommittee Last fall. He had just been grilled by republicans and some democrats on a range of issues. Especially agency proposals that address climate reporting and environmental, social and governance or e. S. G. Investing. Indeed, chair gensler is familiar with the challenges of being a regulator under fire on capitol hill and elsewhere around washington. He served as chair of the Commodity Futures Trading Commission from 2009 to 2014. During that time he led the cftc in enacting rules that reformed the 400 billion swaps market in the wake of the financial crisis. That was a heavy regulatory lift. Now hes undertaking another effort that some see as herculean. The s. E. C. s agenda contains 55 proposals in various stages of the rulemaking process. The proposals address a wide range of issues and market participants. One pending proposal is described as being, quote, related to Investment Advisor conflicts in the use of predicktive analystics, Artificial Intelligence, Machine Learning and similar technologies in connection with certain investor interactions. Were about to hear more from chair gensler on the intersection of Artificial Intelligence and securities regulation. Skeptics say chair genslers agenda is overreaching, overlapping and too aggressive. His supporters say hes pursuing crucial reforms. Chair gensler has said his goal was to improve the efficiency and transparency of Financial Markets and reduce costs for investors. The rise of cryptocurrency has been another challenge for chair gensler in recent weeks. The s. E. C. s division of enforcement has taken several actions against crypto platforms that the s. E. C. Claims are operating as securities exchanges without following Securities Laws. A recent court ruling appears to have set back to some extent the s. E. C. s crackdown on crip toe exchanges crip toe exchanges crypto exchanges. Just as hes familiar with the cut and thrust of washington politics, chair gensler also was steeped in knowledge about Digital Assets, before the he took the s. E. C. Helm in april, 2021. Prior to being nominated by president biden, gensler was professor of Global Economics in management at m. I. T. And served as Senior Advisor to the m. I. T. Media lab Digital Currency initiative. In addition to his experience as a regulator and academic, the chair has served as an assistant secretary of the treasury and a congressional aide. He also has worked for many years he also worked for many years on wall street where he was the partner at goldman sachs. If youve heard chair gensler testify before congress or speak at an s. E. C. Open meeting, you also know that hes a fan of romantic comedies. Sometimes he relates that passion to s. E. C. Proposals. Perhaps hell do so again today. Please join me in giving a warm National Press club welcome to s. E. C. Chair gary gensler. [applause] mr. Gensler thank you so much, mark, for that kind introduction. I didnt know that you wanted me to talk about romcoms today. But i instead have to start with this. As is customary, id like to note that my views are my own, as chair of the securities and Exchange Commission, and im not speaking on behalf of my fellow commissioners or the s. E. C. Staff. Nor, if i can add something to that disclaimer, am i speaking for or by a generative Artificial Intelligence model. [laughter] as the pandemic swept across europe and the United Kingdom my pages are stuck together. Look at that. Maybe the a. I. Model did do this. [laughter] as the pandemic swept across there they go. Thank you. Students packed up their books, they went home. One student at Trinity College in cambridge left for his remote family farm, while in isolation, he continued his work on physics and math. Now, im talking about the 1660s here. And the bubonic plague. It is said that while in confinement, isaac newton contemplated gravity, among other profound insights, including a new form of math, cal khruls calculus. The math innovation. Perhaps the bain of some of your high School Experiences and with journalists in the room, maybe just the bain of your college experiences or your childrens experiences. The math innovation is embedded in so much of life, from pharmacology to finance. Well, in the beginning months of a more recent pandemic, when many 21st century students were packing up their books, going home, like isaac newton did 300plus years earlier, there was a rollout of a new version of a generative a. I. Intelligence model, gpt3. That was already threeplus years ago. Given the Health Crisis that relief may not have gotten as much attention as the subsequent model released, chat gpt4, which was unveiled, yes, on pi day. March 14. Its not when you eat apple and cherry pie. But on march 14, the same day, pi day, two other companies also released a generative a. I. Model. Interestingly, just two days later, chinas largest Search Engine company also released ernie bottom, a chinese bot, a chinese language competitor. A lot of the recent buzz has been around such generative a. I. Models, itkly large particularly large language models. A. I. , though, is much broader than just large language models. I believe its the most Transformative Technology of our time, fully on par with the internet, fully on par with the mass production of automobiles. We dont know for sure where it will all head. But that transformative. Just like calculus, if i can, for some journalists, just like calculus, the math underlying a. I. Is built on a groundwork of many others. Newton built on many earlier mathematicians work. One of them was renee in the 17th century, and those mathematical achievements, yeah, the bane of some other people in this room, that famous linear transformation formula, y equals mx plus b, you might go, wheres heeded, yeah, you can try to figure out, im a bit of a math geek, i like romcoms and i also like to dance. So there you go. [laughter] but you maybe could have figured that out if you were using generative a. I. But similarly a. I. And newtons work on gravity both are based upon data and computational power. So its not just the math. Newton built, of course, on galileos work. Gathered data from other scientists research, his computational power up here, his mind, and his quill pen. Discussions about a. I. , though, go back at least to the mid 20th century. You may remember allenner toking. Imitation game, the enigma code. In 1950 he wrote a seminole paper computing machinery and intelligence opening with, quote, i propose to consider a question, can machines think . End quote. Thats allen turig, 1950. In the 1980s, world chess champion claimed a computer would never beat a grand master. In 1996, he beat i. B. M. s deep blue. Only a year later, carey lost to deep blue. Fast forward 14 years. I. B. M. s watson used brute force computational power to win jeopardy. That was 2011. Watson had been fed more than 200 million pages of documents from dictionaries and novels. Ken jennings, the jeopardy champ at the time, wrote on the screen during the final jeopardy, i for one welcome our new computer overlords. Think about that. That was 12 years ago, ken jennings. In 2016 alpha go using new forms of a. I. To feed it a human World Champion in the complex game of go. Though perhaps not catching as many headlines as these computer versus human experiences and these challenges, or even chat gpt, were seeing a lot of adoption of a. I. Text prescription in prescription in our mobile devices and emails is commonplace. The u. S. Postal service, by god, the u. S. Postal service uses and has been using a. I. To predict the addresses from human hand writing. How else could they write Gary Genslers hand writeing . [laughter] its being used for National Language processing, translation software, radiology, yes, our radiologists can siebert patterns as to see better patterns as to whether that looks like a tumor or not from a. I. , and robotics and a virtual assistant. Im talking about siri and so forth, erica, not that movie her with scarlett johansson. That was a good movie, i should say. [laughter] in finance, a. I. Already is being used for cost centers, account openings, compliance programs, trading algorithms, in fact, this talk may go into a computer that discerns the sentiment of the chair of the s. E. C. What they find, i dont know. And more. Its also due to rapid change in brokerage app. Look, just as newtons work on gravity was based on math, data and computational powers, theyre the three, math, data, computational powers, all three are fueling a. I. Progress today. The math underlaying a. I. Has moved from brute force computing, like what happened in that jeopardy or what happened to gary, from brute force Computing Power to whats called new orleans networks to Neural Networks to deep learning. What that math is may move on. Having layers of individual nodes or, yes, neurons. The parameters, to speak math, parameters are these coefficients like the m and the b back to that. But the parameters, the coefficients relating the connections between the neurons no longer number in the hundreds, they number in the thousands, millions or even billions. Decart would have been pretty impressed. Collectively the models are being optimized to discern patterns in data and making predictions. We also, though, are living in an era of exponential data growth. Datas being drawn, of course, from social media footprints, our shopping, our spending. But more than that, its also being drawn by an explosion of sensors. Cell phones, of course. Fitness devices. Telematics in the cars, the actual sensors in the cars. Appliances, cameras, and other socalled internet or thing sensors. I recently installed a new heating and air conditioning unit in my home and i was asked for my wifi code so that could tie into the internet. And someone i guess has another bit of information on the chair of the s. E. C. I turned it off. [laughter] i just want to say for the record, its not you know. There are bound to be more sensors than humans in the united states. We also see in the early stages of whats called multimodal Artificial Intelligence systems. Thats where you combine multiple types of data. Video, audio, speech, images, text. And you try to make even more accurate determinations when you have multimode ales coming in at you multimode ales coming multimode ales coming in at you. The ability to increasingly exponentially have more computational power. So its math, its data, computational power. This isnt a forward direction. Its like wily to its likely to continue. A. I. Open its up tremendous opportunities for humanity. From health care to science to finance. As machines take on pattern recognition, particularly when done at scale, this can create great efficiencies across the economy. In finance theres greater financial inclusion, enhanced user experience. When newton pondered the laws of gravity, he thought in the micro, the apple. He thought about the macro, the cosmos. So if i could be inspired by newton a little bit and in that vane talk about a. I. s advances, what they might mean in the micro to each of us as individuals and the macro to the broader economy and society. So starting on the micro. What i might call narrow casting as opposed to broadcasting for all you journalists. Narrow casting. Today a. I. Models provide an increasing ability to make predictions about each and every one of us as an individual. This growing capability facilitates being able to differentially communicate to each of us and to do so efficiently at scale. How might we respond to an individualized communication . How might we respond to an individualized Product Offering . How might we respond to an individualized pricing . Communication, products, pricing, can really tailor and individualize, in other words, narrow casting. Quite a difference from turigs day at the dawn of network broadcasting. On a daily basis already we receive messages from a. I. Recommender systems that are considering how we might, individuals, response to the props. The models have been developed to assist in making decisions already about who gets jobs, who gets loans, who gets credit. Who might get the entry into a school. Who might even get a health care outcome, where that m. R. I. Is going to be for us or somebody else. And thats just to name a few. This raises a host of issues, not necessarily new to a. I. , but accentuated it. One is, when i was at m. I. T. , when studying Artificial Intelligence, we used to talk about explainability, bias and what the computer scientists called robustness, you might call accuracy. A. I. Models decisions and outcomes often are unexplainable. Part of this is the inherent models themselves, the math. Remember, that math i was trying to skraoeufplt its nonlineal describe. Its nonlinear. Its hyperdimensional. Meaning its up to thousands or even billions of dimensions. And thus its hard to explain. But its also dynamic. It learns from new data and from the models use. So its not static. Thus the insights that come out of such models by design are inherently challenging to interpret in terpgs of accessibility for us humans. Why does the model, for instance, qualify one of us for a loan and somebody else not for the loan . A. I. Also makes it more difficult to ensure for fairness. The outcomes of its predictive algorithms may be based on data reflecting historical biases as well as features that may be proxies for protected characteristics, right in the data itself. Further, the challenges of explainability may mask underlying systemic racism and bias right in the predictive data analystics analytics. The availability of these models to predict doesnt mean theyre always accurate either. They might be predicting based upon somely a ten tent some latent feature. Does the model look at a photo of a dog and say its a dog or is it a cat . A couple years ago, this was back in 2019, my identical twin brother, rob, hands me his iphone and he says, what do you see . I say, i see a text message. He says, dang, he said, that was set on facial recognition for rob gensler, not gary gensler. You might say, thats ok. Identical twin. But these challenges of Data Analytics are not new. In the late 1960s, early 1970s, congress weighed in. Fair credit reporting act, fair housing act, equal credit opportunity act, all in part with similar issues. Making sure that people had explainable and fair and unbiased Consumer Credit in housing and the like. As advisors and brokers incorporate these technologies and their services, the advice and recommendations they offer, whether or not based on a. I. , must be in the be