A Conversation with Alphabet Chairman John Hennessy

hi everyone uh good morning thank you for joining us today my name is Isha chakraborthy I'm a current senior here I'm at Caltech I'm studying computer science and minoring in information and data science we are so pleased to have all of you gather for the event today both everyone here in person the Caltech students and postdocs as well as everyone who's able to join us online

and other members of the Caltech community so today's event is the fourth in a series called the Breakthrough insights a conversation with the world's most influential Business Leaders so we are incredibly honored and excited to have John Hennessy here today the chairman of alphabet and leading the conversation today will be caltech's president Thomas Rosenbaum so I now welcome president Rosenbaum and John Hennessy to the stage

great thank you Usher and welcome to all it's really a great pleasure to be here and I'm particularly delighted to be able to introduce John Hennessy I'm going to then ask John a few questions to get the conversation going and then turn it over to you as former president of Stanford University a champion of multi-disciplinary Education in the Arts chair of the board of alphabet founder

of companies member of the national academies of Science and Engineering John Hennessy has contributed to Science and Technology Academia and Industry research and education in Myriad ways he may be most famous quote for pioneering his systemic quantitative approach to the design and evaluation of computer architectures with enduring impact on the micro processor industry in the words of the citation for the 2017 touring award which you

receive jointly with David Patterson in the early 1980s prevailing wisdom was that processors work best with fewer instructions regardless of how complex those instructions were as John said in his touring award lecture microcomputers were competing on crazy things like here's my new instruction to do this kind of thing rather than saying here's a set of benchmarks and my machine is faster than your machine Hennessy's work

challenged that assumption proposing that Computing would be more efficient with larger but simpler instruction sets in 1981 he drew together researchers to focus on technology known as risk which revolutionized Computing by increasing performance while reducing costs in 1983 building on this work Hennessy Stanford team developed a prototype chip called mips microprocessor without interlock pipeline stages commercialized this product he co-founded mips computer systems in 1984 and

served as its Chief scientist for eight years and Chief Architect for six more that mips was later acquired by silicon graphics these successful traits already emerged in high school so we'll go way back where John won a science fair prize for an automated tic-tac-toe machine built with Surplus electrical relays he recounts it really got me excited about approaching challenging technical problems working out Solutions and doing

things that were driven by my own initiative John joined Stanford University's faculty in 1977 as assistant professor of electrical engineering and spent almost the entirety of his career there serving as chair of computer science dean of the School of Engineering Provost and is president from 2000 to 2016. in 2017 he initiated the night Tennessee Scholars Program the largest fully endowed graduate level scholarship program Scholars receive

full funding to pursue a graduate or professional degree in any of Stanford's seven schools with the goal of preparing leaders to make a positive impact on the world a fine summary John of your own approach and contributions thank you Tom oh well it's just such a pleasure to have to have you here that was an introduction that my father would be proud of and my mother

would believe so oh but you can Count Me among the Believers no problem so growing up in Huntington Long Island you were a tinkerer from way back was this something that your parents encouraged since you mentioned them yeah my father was an engineer and worked in the Aerospace industry remember this is around the time that Sputnik went up and there was a panic in the U.S

that we had fallen behind on various key Technologies and um so he encouraged that and I did the usual things that young people do blowing up things and all kinds of science experiments and things like that but I got interested in Computing which in those days there were no courses you joined the computer club which meant you had access to a teletype connected to a remote

far away computer you wrote in basic or maybe Fortran and you put your programs on paper tape but it was a great way to get introduced to the field and begin to to Really build my Fascination and love for the field of computing so you knew from early on that Computing was the direction you wanted to go I did um when believe it or not when

I was an undergraduate you could not major in computer science anywhere in the United States yet because the field was still you could graduate programs had emerged but no undergraduate programs so I became a double e major and kind of focused on Computing as my core and then went and did my PhD in computer science right so you went to college at Villanova a very different

model than Stanford very different did you take things from that part of your experience that you wanted to translate to Stanford so there were two things that really shaped my undergraduate experience one was an opportunity to work in undergraduate research which really inspired me to go do my PhD and the other was I taught a small problem-solving session for students taking their first programming course and

that really Built My Love of teaching and I decided I decided when I applied to a PhD that I wanted to be an academic at the beginning and I probably would have stayed in academic except other things intervened and came along and opportunities to take technology out into industry became really important right so you moved between academic Academia and industry very fluidly and very early on

that didn't happen in fact the two two companies that had made the largest investments in experiments both canceled their projects and decided not to turn into products so Along came a famous Pioneer in the computer industry and said to me you need to go start a company because this technology is disruptive it's going to obsolete their existing product lines and they're reluctant to upset if you

really believe in it you have to do it so I had had some earlier startup experience in fact my first startup experience was in a Caltech spin out a company called silicon compilers that Carver Mead started during the Mead Conway the LSI revolution of which of course my work was really enabled by that Revolution that let system builders get their hands on Silicon which up until

that time had not happened um so I you know I decided to go out and start this and I spent a year and a half full time in a couple years part-time after that but in the end I missed students too much so that you came back to the university to be with students that's good and when you talk to students now and they're thinking about

different directions what do you tell them you know I tell them the great thing about being an academic is it's a terrific lifelong career and you have to decide what really motivates you what excites you what because you I tell young people you really want to be committed you want to be eager to get into work and do whatever you're going to do whether it's teaching

a course doing your research or building a product that other people are going to use and are going to enable them to do things so find out which of those things gives you the most inspiration and figure out how to do it yeah great advice for for any direction so you talked about being part of the Mead Conway Revolution but when you were working on your

computer architectures they really flew against the wisdom of the day I mean your collaborator David Patterson at Berkeley called the ideas that two of you were advancing quote controversial and even heretical um so so what was it like was it frightening thank God Dave Patterson was there because otherwise it would have been Me Against the World um and at least there were two of us that

had similar kinds of results and we knew that IBM had done these experiments with the technology so even though they decided initially not initially not to commercialize it they did subsequently but initially uh they didn't we knew that there was other confirmation of the ideas um but there was a famous uh incident where we were on a panel and Dave and I were on the panel

and there was a person from another company who was a naysayer and um one of the the moderator of the panel says to the naysayer um Hennessy is just raised a million dollars to go develop this technology but you say it's not going to work what should he do and the person says he should take the money and go to South America so I didn't go

to South America and the naysayers project got actually canceled by the company who was working so it all worked out in the end but it was was it was hard to convince people and yeah they're really while while uh Carver Mead had kind of urged the creation of foundries things like companies like tsmc and right and fabulous semiconductor companies which didn't own Fabs it hadn't yet

happened and that was one of the hardest struggles was building a business model that worked and I get one of the insights I've gained of this over time and I've told young people that are thinking about startups and I'm counseling them I said look either you should do new technology or you should do a new business model don't try to do both because that just increases

your risk dramatically and will lead to lots of failures that but that was hard we had to sweat things out it took longer than we it should have because if The Foundry business had really been vibrant the way it is today it would have been much faster so what prepared you for the new business model part of this the technology side of it is more straightforward

in some ways yeah and the first thing we did was we went and hired a great technical team I mean really terrific top-notch A-Plus technical people who were excited about the idea on the papers we had written um I didn't I I didn't know anything about running a business I knew nothing I could not read a balance sheet I didn't know what a gross margin was

I mean really I didn't know anything and the result of that was we made a bunch of mistakes um and we had to change CEOs at one point we had to go through lots of difficult hard things and eventually we got a CEO who built a team that knew how to do those things much better and could go build the relationships we needed with potential semiconductor

foundries and so after when I came back to the university I said you know we're going to have a lot more entrepreneurs at Stanford we should make sure that we're providing them at least the basics so if they're going to go out and do something start a company or even argue within their existing company for a new product they should be able to at least pitch

their ideas in the language that the business side of the of the of the institution of the company will really understand so interesting you put it that way because I think language that the disconnect often you you have the same ideas but if you're not saying the way people connect to it and it's not something this isn't rocket science it's just you have to know you

have to know the facts we went out we thought I thought but we don't why do we need sales and marketing we have such a great product people won't come to our doors knock on our doors and buy it I really thought that I really thought that so we learned we learned differently we did need sales and marketing now you weren't alone you know I think

caltech's view for many years was if you didn't know about us it was your problem I'm sorry [Laughter] a lot in the news today talking about next steps in terms of architectures and and business models of course about generative AI we've spent time today actually talking about it as well if you want to give us your your view of where we're going where we're at where

we're going yeah uh this is a phenomenal time of change I've never seen a technology move as fast and I've seen I've seen the microprocessor be created I've seen the personal computer be created I've seen the internet be created I've seen the worldwide web being created the rate at which machine learning based AI is moving is faster than any of those things and the potential impact

I think we used to say well there's the internet and all these other things and AI won't be as important as that and I think now things have completely flipped and people think the the work the progress on machine learning based AI is going to probably be the most important thing that's ever happened in Computing which has this incredible history over the last 50 60 years

70 years so I think and the other thing that's amazing is we're really seeing a kind of emergent Behavior if I look at the Systems computer systems we've built over time really complicated ones even I don't think anybody who implemented those systems expected the the system would do something that was beyond their expectations these generative AI systems these large language models do things which are beyond

the expectation that people have and I think that's a real change point and it's a real difference and maybe it's a combination of getting to sufficient scale because these these models now have hundreds of billions of parameters in them and having sufficient training data that they really have embedded lots of information in them and that's those are the two things that really and and the computational

power to do that too I mean these are these are training on the fastest computers in the world for three months to a year yeah full-time that's the only thing it's doing it's not doing any other problem it's just doing that one pro that one training set for that time so phenomenal amounts of computer uh investment in doing it but it's a phenomenal time so which

Fields do you think are going to be most radically transformed I think you probably should invert the question which field will not be dramatically transformed robotics is changing at an incredible rate medicine biomedicine is changing at incredible drug discoveries changing at an incredible rate um materials look at the work going on and using machine learning and materials I think lots of people focused on trying to

find a catalytic methods for example to remove CO2 from the atmosphere and now using machine learning kind of thing I ran into just in this past week I ran into two people working on solving computational fluid dynamics with machine learning it's kind of Boom it's moving so fast partly because we're getting a lot closer to to artificial general intelligence we're not there yet we're probably still

10 years away but but everybody used to think we were 20 30 40 years away I think people think now 10 years it could happen so that touches on the ethical side of this and in an interview and wired you had a quote which I thought was wonderful about training our students to contribute to society you said if you think about the young group of undergraduates

coming in a bunch of them are going to be doctors a bunch are going to become Business Leaders a bunch are going to be Tech leaders a bunch will go into politics in all those areas understanding ethical conduct and the consequences of important decisions seem absolutely crucial to me so are we grappling with this appropriately probably not sufficiently I think first of all this technology has

exceeded what we thought it would be able to do I think for lots of people chat GPT can pass the Turing test for a little not maybe not an expert in the field but for lots of people it would pass the Turing test um so it's got some sense of intelligence at some level um whatever exactly intelligence means which of course that is turing's original argument

was you can't Define intelligence except by not being able to distinguish it from the way a human acts so I think the tool itself is neither good or bad but it can be used for lots of negative consequences and I my feeling about this is we have to prepare young people while they're in college to deal with these ethical kinds of dilemmas by helping them sort

through what their values are what their goals are what the limits are because when they are doing it in real time in an industrial setting things move very fast and there isn't a lot of time and if you look at our biggest ethical failures they're all characterized mostly characterized by slippery slope in general people don't do something that was completely ethically wrong there are bad people

in the world but most good people don't do that what they do is they do something that's a little bit down the slope and then they take another step and they take another step and before you know it it's an avalanche yeah I I I'm struck as well by the the notion of um Advanced preparation because of the pace of change and I you know there

have been calls for slowing things down is that a practical solution from your point of view I don't think it's going to I well first of all I don't think it's going to happen because it's for better or worse it's an intensely competitive environment this is intensely competitive the big companies are really competing really hard against one another that makes it very hard unless everybody agree

and then you know how do you ensure how do you as as my friend my former colleague used to say George Schultz used to say trust but verify yeah so I don't know how we build a trust and verify uh for the industry I think more likely that the industry might get together and help uh endow and and operate universities to figure out these ethical issues

and help support universities to be the neutral broker that talks about how we try to ensure these Technologies get used for good uh rather than evil but we've got the other problem Thomas we've got this technology that's obviously going to do a lot of good I mean you look at what Alpha fold has done for people in terms of protein folding and its use then the

question becomes we can't take a technology that has so many good opportunities and hold it back we've got to figure out how to constrain the negative opportunities the negative consequences and I like very much your notion of a special place for universities because we are the kind of Institutions that could contribute in this fashion and we're the long universities are the long-term thinkers and they have

the diversity of of intellectual interests that can let them really focus on this right and and the humanistic component is so important I agree in fact I that that leads I have another uh uh incident that you recount in in your book leading matters where you talk about the violent illness ishak Perlman coming to Stanford and in The Green Room Perma and greeted you saying Mr

President Stanford is a great University but you have terrible performance facilities even before that you you had seen support of the Arts as a critical aspect of an education um so how do you see this intersection between the Arts and Humanities and Science and Technology we live in a world that is really about human beings and the quality of their lives and that's so deeply enmeshed

with what we are as people and the role that the humanities play history language culture but also the special role that the Arts play the Arts uh have a way of dealing with things like ambiguity that and science is a much harder we don't like ambiguity in science we try to stomp it out right but in the Arts ambiguity is is a part of what makes

the Arts Special similarly their ability to talk across cultural boundaries and to bring together people to gain an appreciation of different cultures so I think it is a foundational thing that we want to be part of a student's education and their experience um and if you if you came up to our our building where Knight Hennessy Scholars are you'd see we we're building this very diverse

art collection from around the world to really help our Scholars engage in these kinds of conversations yeah so I love the idea about ambiguity and also what your um speaking of which is empathy and generating empathy through cultural appreciation through stepping outside of your own experience and yeah the Arts I agree have a special role that way um so my last question at the end of

the story though yeah yeah please so that was a great story and I I got embarrassed by Itzhak Perlman right about the quality of our facilities um but that was a great story to tell a donor about why we needed money for any Performing Arts facility which works so we now have Bing Concert Hall that has a great performance facility for it and in South Portland

has come back and played it and loved it so so this Stanford campus was transformed into your leadership I mean not only the physical campus but of course the intellectual portion when you look at it what rings out to you so the the so a couple things got changed that really um the thing that we did that gave us the most pleasure and received the most

alumni support was making a big change in financial aid we basically eliminated tuition payments for students from lower income families and that I I never got so many great alumni letters the alumni thought it was terrific because we were going to ensure that Stanford would be open to the very best students irrespective of their ability to pay but the other thing was what we were able

to do we really rebuild the art side of Campus um and we built some really wonderful facilities and made the Arts a much more vibrant uh part of the campus and I'd say you know we we had this view that there was going to be a lot more multi-disciplinary collaboration in the academy and people working across boundaries and we started by building one building right on

the boundary between across the street is biology and chemistry across the street in the other direction is engineering and it backs on to the medical school and we built a building that had faculty from 20 different departments in it and that was really the beginning of a real change in how the university thought about boundaries and collaborations and these sorts of things which I think um

today is one of the strengths today so we have a brand new school of sustainability for example that um that is a highly interdisciplinary and includes people who are working on environmental law and environmental ethics as well as people working on science and technology to Halt the danger of climate change yeah I'm just going to take all those disciplines it's going to take all those disciplines

it's the hardest problem we've ever really had to solve I suspect yeah uh I have one last question before we turn it over to the audience and that you've confessed to be an Aficionado of uh 19th Century door stoppers reading Victor Hugo's Les Miserables and Charles Dickens A Tale of Two Cities at least a few times so so I had two questions one was what what

do you find when you return to these novels and the second is what's on your night table now ah um so I find that when I come back to read various books my depth of appreciation is much greater and there are some authors which I tried to read when I was in high school or I was asked to read when I was in high school we're

told when I was in high school Henry James for example which I I didn't really appreciate in high school it it requires more of a life experience to really appreciate and some of George eliot's books are similar middle marches certainly that way Mill March is a great book but if you read it when you're 15 or 16 you really won't get it when you read it

when you're older um but I I find real that that incredible life lesson um I think really comes back and that's one of the really great things about reading and about great works of literature what's in my I'm just I just finished um a book by one of my colleagues Abraham berghis a covenant of water he's a he's a faculty member in the medical school that

wrote a bestseller called cutting for stone and his books are a mix of their novels they have a mystery component but they also are deeply embedded in a culture of the place in this case uh Caroline in Southern India remarkable if you have a remarkable Community um and he he stitches a story over multiple Generations there that's really a joy to read and I love great

books that comes through so thank you John it's wonderful that you're willing to share your insights and your perspectives and and I'm sure we will learn more in the questions so please if you have questions just raise your hand yes back there um thank you for your amazing talk it's been very excited and hear about the you know the night Hennessy program and whatnot so I

have um I have a question about the relationship between Academia and um well the relationship between different fields and then the relationship between Academia and uh I guess commercializing startups especially with hard tech there's often that value of death in terms of translating academic research into commercializable products what are some of your thoughts on the ways that we can you know further bridge that Gap and

then tangentially within you know PHD programs we're often encouraged to really specialize but especially for many of us not aiming to go into Academia where perhaps breadth alongside depth would be helpful what do you think are some ways that maybe Universe PHD programs in general can maybe take on some of that Spirit of the night Hennessy and encouraging interdisciplinariness uh yeah so I I think uh

there are there are challenges for some technologies going from the invention in the lab which often happens in a university particularly for some of the most revolutionary Technologies uh companies are very good at incremental steps they're they have a much harder time with the big uh big revolutions I think the challenge in a lot of those I mean you take the in Green Tech for example

um you know you may be looking at you have a new battery company you're probably looking at 15 years of development sales before you get to the point where you're actually really selling product at some reasonable volume as opposed to giving out samples of things so we've had we're going to have to reform our funding systems to align with that which I think is a better

solution than trying to figure out how the universities could fund these projects longer when the the work that needs to be done no longer looks like the research it looks really like development and it's not a good use of graduate student or university researchers time so we're going to have to figure out and I think we see some things now I mean there are some new

energy funds that are out there that I think understand the importance of the climate change problem and they'll have more patience in working that with respect to phds some phds are going to go into Academia and we should prepare them to be to be great academics some are not many are not these days uh particularly in stem Fields the majority don't go into Academia so you

know we need to slightly broaden their education and prepare them to do that and get out there and we probably all of us could work on not letting PhD times grow longer than they've already grown they've grown too long it's bad and and I think it's terrible and we should try to get that time back to where it once was four to five years rather than

five or six or maybe even seven on occasion so yeah yes over there I think a microphone is coming your way yeah thank you very much for being here to speak to us a little bit of context I'm a mechanical engineer by trade I went into that field because it had so much breadth and the capacity to help people potentially that needed it and as I've

learned I've learned how to use the tools of the trade well and I understand what they do when I use them if you give me a hammer I know how to hammer a nail into a piece of wood and I know how not to hit my hand when I do that but if you give me chat GPT I don't necessarily know how to use that tool

to impact Society in a way that's necessarily good for human flourishing right and maybe not just like local human flourishing here in the United States but globally how can how can I be thinking about developing these new technologies for the sake of the benefit of humankind and then on top of that what what direction Academia industry starting companies do you think is the most effective at

promoting the ethics of new technologies yeah it's a good question I think you should think of chat GPT any of these other large language models you should think of them as a natural language interface good very good at manipulating words into various forms but that that's only the beginning of what the machine learning Revolution is going to do it's going to help you design new products

in mechanical engineering it's going to help make those products easier for people to use partly through speech or language interface but through other other kinds of interfaces as well which will get better because the systems will be able to Intuit what a user wants to do so really don't don't I mean I know generative AI is taking over everybody's excited about it it's amazing what it

can do but think of this as only one small part of it and I think what's going to have to happen it's you think of an earlier time I suppose that the Department of Mathematics has said anything that has to do with math belongs to mathematics and physics can't do the math part and chemistry can't do it mechanical engineering can't do it the world would never

work that way that's exactly what's going to happen with machine learning it's going to have lots of people going to be using it in their disciplines as a tool in the same way they've used math or computation as a tool in their discipline and that's how we're going to get really great new things built that serve Humanity well right so as Isha mentioned we have a

number of people tuning in online and I believe you are some questions from our online viewers hi um so I'm asking this question on behalf of Vincent ton who's the program director at the student faculty programs so he says good afternoon Dr Hennessey thank you for joining us at Caltech uh you mentioned how undergraduate research greatly influenced your post Villanova experience as Caltech celebrates 45 years

of the summer undergraduate research fellowship or otherwise known as our serve program what advice can you pass on to these students the advice is simple you are at a great research Institution take advantage of the fact that you're at a great research institution if you don't take advantage of that you're missing some of the joy and the experience and the growth of being in a place

like Caltech so absolutely take advantage of it you'll have Great Courses but go get involved in some research couldn't couldn't agree more I figured that well you create Community as well so but of course we're biased observers that's okay yes right in the middle there sorry the the microphone is coming okay you mentioned that Alpha fold was one of the big game changers of biosciences and

that because of computation drug discoveries like a very um possible way to further impact the world currently right now it's like a decade-long effort with a billion dollars spent into making one drug if it like succeeds through all the faces of the drug trials so how do you see computation potentially changing drug Discovery for the better yeah I mean I'm I'm not an expert on this

but having talked to my colleagues I think uh the the we have a couple problems in current drug discovery part of it it's a bit of a needle in a haystack problem of searching through things well computers are much better at searching through possibilities than people are we have the problem that the gap between finding a Discovery at lab and actually getting through clinical trials is

a very difficult one and long and very expensive suppose we could speed that up one area which I've I've seen some colleagues working on that's very promising is to find alternative uses for drugs that have already gone through clinical trials so we know they're not toxic and if we can find good uses for those drugs to solve other kinds of human problems we can probably get

them out there much more efficiently and cheaply than developing a whole new drug all right thank you yes over there I'm happy to provide good exercise to the microphone carrier hi thanks so much for coming um I'm a grad student here in chemical engineering and I was actually Stanford class of 2015. so you're my president and thank you so much for everything you did while I

was there um I while I was at Stanford during your time there was this massive increase in computer science majors and it surpassed hum bio human biology as the most popular major and you saw the number of BS degrees like you know shoot up and the number of Bas go down and I think of course like it's natural that during this time we would see so

many more computer science Majors the same thing has happened here at Caltech and I'm curious you know as a chemical engineer it's obviously made me a little sad to see like fewer chemies right and like and of course in the Arts as well and in the humanities I think I share your um belief in the value of those things and in a liberal arts education and

I'm curious how much does a university allow this to happen how much does it fight back and try and like you know preserve an element of the diversity of Interest how does the University think about doing that um I'm curious what you think yeah no I think it is it is a concern I think we've always most of the uh top institutions have kind of said

well students make their mind up of where they go and we try to make sure they have a great educational experience but I think we do worry about it I think in some fields we can see a response to it so building out a lot more quantitative uh computer-based uses in particular fields in the social sciences in particular which is happening so fast in the social

sciences so I think that will draw some students who really they really want to be in the economics or political science and once once the the curricula will support people who can do much more computationally oriented work in that world I think that will that will help I mean I think we need to keep the humanities as part of the core base of the education and

something that our experienced because we're going to students from Caltech are going to go out and be in important roles around the world whether it be in government or non-profits or Academia or in companies and we need to prepare them to be the kind of leaders that we'd like to have in the world leading leading us forward nice thank you so Isha maybe we could take

a couple more questions from the online audience all right so um this is this is a question on behalf of Ben picker who's the Caltech student class of 2012. our Celtic Alum I guess um so he asks what are the most important things you think our government needs to focus on in order to maximize productivity in terms of AI policy he notes that between China Europe

and the us we're seeing different pros and cons of philosophies of government and their fluidity and adaptability strongly impacts our ability to have competitive markets and also our ability to regulate AI to ensure that it's good for society he's asking you what are the most important things that the U.S needs to focus on for the next five years to remain to ensure that we remain productive

but also secure as we develop AI technology sorry there's a long question yeah I think the observation that uh China the U.S and and Europe think very differently about these problems is absolutely accurate I think we need to get a balance so let me let me start with the first thing we need to do is and this is going to be extremely difficult we need to

agree internationally on the limits of use of AI with respect to military use we must do this if we do not do this it will come back to haunt all of us because people will develop weapons using AI technology and the alternative will be everybody will be forced to follow suit and it will be it will be a terrible thing for for humans on this planet

so we're going to have to do something about that to make sure humans in the loop but that also leads to another key insight if we say there are humans in the loop that make decisions the same thing can apply to lots of important AI applications so we're using artificial intelligence to amplify the ability of somebody who's a professional so take the case of a doctor

looking at x-rays for example great the AI system is going to look at this but I don't want to talk to an AI system have a voice say oh I had to tell you Mr Hennessy you have something terrible that's not the experience I want right I want to talk to a person now that person may engage with the AI system look at it make sure

they really believe it and they understand what the results are that's the way I think we ought to be thinking about these Technologies working so that they're supporting humans they're they've got to focus on human betterment and we're going to have to agree how to do that and how to get the Technologies to work in that direction the whole set of other problems deep fakes all

these kinds of problems we're going to have to find solutions to those in strengthening our democracy and our civil systems so that it can withstand that because they're going to there's nothing you could do to prevent those from occurring we're going to have to think about how do we build systems that are strong enough to withstand them good question yes over there yes sorry it's indirect

uh thank you also for coming today um I just have a question of about how you mentioned the ethical issues that come from that like the rapid movement of AI and with AI becoming so Advanced so quickly you said how it's a responsibility on University specifically to put things in place so that students who are going to graduate and go into these fields know how to

ethically decide things before they start moving towards that downward slope but a bit like to get more specific in that do you have any actual like ideas of how a university would go on to implement like teaching especially because in the current world we have like as AI gets um more advanced we as a society even in general before AI have been getting more null and

like dull to the terrible things happening since we see it happen every day all the time and so when you when now everyone is going to have access to such powerful technology terrible things could easily happen and so how do how does the University actually put things in place so that that doesn't occur well I think the University's job is to educate its students to be

the kinds of responsible leaders we'd like to have that's that's the primary role the educational role that the university has and you know I think one way to do that one way we've we've tried to do that at Stanford is by having an Ethics in the major requirement so it's not just an arbitrary ethics requirement it's an Ethics in the major requirements so you're in computer

science it uses examples drawn from computer science but again when we educate people we're not necessarily picking the exact sample you're going to see 10 years later when you're out in the field we're trying to get you to understand the general problems how to set your values how to be aware of slippery slopes issues like that there's a whole other role for the University which is

to be the place that thinks about AI ethics in a deep way that puts together technologists with ethicus and thinks about how would you build a set of principles that encouraged good use of AI and discouraged bad uses of AI and that's an important role for the universities to play we have the long-term thinkers we've got the set of disciplines to do it we should be

the ones who think about this problem right in the front here hi thank you Jen for coming so I'm on uh frosh here at Caltech studying computer science but I plan to go into research grad school and you were mentioning how fast the field is changing and that's kind of intimidating how do you go about not thinking oh the work I'm working on now right or

the work I do my PhD in is going to be obsolete as soon as I graduate and how do you deal with just such an ever-changing field you should think of the work we all do when we're working in research I like to use an analogy of um we're we're building a foundation that other people are going to build on and that's the way science and

research in any Field Works we build a foundation and the foundation I want to build is one that people who come after me have to build on top of it because I built a foundation that was Secure and strong and had good insights in it and important research accomplishments and it will get obsolete the work some of the work certainly some of the work I've done

has been obsoleted by progress that's been made since then um and that's that's perfectly fine because we built on top of that and used that as a as a foundation for things that came after yeah and if I could just add briefly there's a certain irony the discourse today is we should be training students to work in particular fields or solve particular problems but actually as

you point out the world is changing very fast so we should be training students to be flexible to know what uh primary sources are and how to use them to solve problems generally and then the world can change under your feet but you will adapt and you'll still be able to have modes of thought that make contributions so you you should retain your confidence yeah we

should we train you to learn and learn new things we train you to be lifelong Learners that's what we want to do and in graduate school we train you at being comfortable walking into a field that you don't know a lot about learning what you need to learn and make a contribution so that that's the goal so Isha do you wanna so for efficiency sake why

don't you take two questions and then well I can do that I'll save you another trip sorry about that um this is actually a question that's kind of related to what we were just talking about so this is um a question I'm asking on behalf of Anna Liu who is a PhD student and was a Caltech student and the class of 2015. so she asks that

many phds who've moved into industry often see this as a one-way Street and so as someone who's moved quite fluidly between Academia and Industry what advice do you have for PhD students currently in Industry who are still interested in working working in Academia and additionally what are ways that we can contribute without following the 10-year Professor track yeah you're right it's it's sometimes a one-way street

it's harder to move back it's harder to move back into the university once you've moved in Industry I think a couple things make it possible figure out how you can publish your work so that you get visibility out there and you get recognition for your work about a different role in moving back if you've been very successful in Industry maybe you can move back into a

different role in the University not necessarily A tenure line role but find a find a different way that you can come back into the university maybe you'll come back lots of lots of people who've been in Industry come back and teach for example at our business school and they love teaching in the business school because they teach these brilliant young mbas it's a different kind of

role but I think there are ways to do that but if you want to move back into a research position keep publishing keep publishing so your work is visible and I'll take I'll do another question from the zoom so this is from an anonymous question asked asking how would you advise educational institutions to factor in new AI tools into their curriculum and teaching methodologies to stay

current well I think we'll get we um when any any technology comes along that revolutionizes field so much we have to rethink about how we teach the curriculum and how we modify that and the way that's going to happen I think is we're going to hire a new group of young faculty that are find this technology to be the hot thing the tool they want to

use and they'll be the ones who help us redesign the curriculum and reorganize the curriculum okay so I think we have time for one more question on the aisle there hello uh so I'm in the Aerospace major and I study Hypersonic flows so these kind of ethical questions are at the Forefront of my mind and something that is very Troublesome is there's kind of dichotomies so

if I go work at St Lawrence Livermore I could work on Fusion research which on one hand could be revolutionary for clean energy on the other hand you could use it to make nuclear bombs and then this could even be true with somewhere like Google like maybe they're making huge things that help everyone but then you have people like Jeffrey Hinton who's leaving due to ethical

concerns so for people who want to go into industry do you have any advice for how we can kind of navigate these dichotomies of good and bad or seemingly both I think I think it's the case that all many technologies have have good uses as well as potentially negative uses um the goal has to be for the for the company the leadership which includes the technical

leadership to try to think hard about how they focus on the good uses to minimize and and and what they do to minimize the possible bad uses how can they constrain them uh how can they ensure that this doesn't happen and I I think you know the basis of your question in the AI field this is going to be difficult it is not going to be

easy it's going to be hard to figure out how we prevent negative uses um and the the challenge we face is this technology can do so much good and accelerate so many fields how are we going to make sure that those the balance here is heavily in favor of those things and we minimize the weight on the other side of the scale that's the challenge we

face hard challenge maybe not as hard as climate change but maybe just right after climate change what a wonderful summary I think to end on and a challenge to the really bright creative young people here to lead us forward so thank you John thank you all there is no sign saying rain so that means we are having a reception out on the patio and please come

and be able to meet John Hennessy there thank you thank you [Applause]