AI for Good
good morning and welcome everybody here this is day two of the tekken society uh conference as many of you know we are pioneering and trying to innovate on a new format which is a hybrid format both in person and remote we think that that hopefully will open up attendance further to people that aren't even in the area locally plus have a good quality uh discussion here as you know with the goals of the conference we're trying to really highlight where can technology be used for good and we're trying to think about with all of that what are the tech lash issues data privacy issues anti-trust issues future work issues and how do you navigate the balance between the two of those uh issues um we had a great kickoff yesterday with juan enriquez and hopefully most of you were were there a lot of interesting thoughts about life sciences about the genome project and then getting into a much broader discussion about technology and society and how do you kind of create better opportunities some good kind of reminders from him about how to think about problems and be really thoughtful about collecting the context the information before moving ahead today's conference we're gonna have three uh topics so ai for good is gonna be the first topic which couldn't be again a more timely one with explosion of data and the ability to analyze it second topic is gonna be global innovation so innovations from around the world and a reminder there a lot of leapfrog innovation is happening outside the u.s and that's why looking at business models product services etc outside the us matters so much and then a third area third topic is going to be green tech and green tech how can you use technology to help in one of the biggest areas of concern in society today climate change sustainability uh etc if i were to give you advice or make a request as you listen to today's sessions two things i'd ask one is to get in the ring on the issues and what i mean by that is hear people out in what they have to to say on all sides thinking about what technology really can do in areas like health care and transportation etc and then think about what are some of the challenges um there it's very easy to get absolutist on positions to get extreme on one side or the other which starts to kind of create a question about do you really have all of the facts there and then the second issue and as a request is put yourself in the role of a leader you know this is one of the defining things about being at a business school and focusing on leadership leaders have to take decisions they have to get a group of people together and they have to go execute against things so just understanding an issue and taking a position on an issue isn't enough it's one of the necessary parts but it's not all of leadership so think about what the leadership imperative would be as you listen to these things and i always find this kind of two questions that define leadership what would you do so not what's your point of view but what would you actually go do if you're in the position of a leader and what does success look like what are you ultimately trying to achieve and i find those are reconciling questions to a lot of the kind of disparate views that that exist so first session that's going to be up is ai for good and we couldn't have a better moderator than heather caruso heather is our assistant dean for diversity equity and inclusion her research she's a faculty member here she's an adjunct assistant professor in the management and organizations area her research is focused very much on collaboration on team dynamics and on leadership one of the frameworks that she has brought to the school that i think is super helpful mavic i use in my class is the echo framework that really gets people to think about are you really taking on information are you reflective about it are you humble and how you you execute on things etc so it's been a huge help she's got a great academic background on top of everything stanford undergrad and phd at harvard so heather let me turn over to you and uh and welcome thank you so much hi everybody um it's a it's a real pleasure to be with you and i really want to thank you for for attending this session of the tekken society conference i also want to take a moment to thank terry kramer and the whole eastern center team for giving us this space to have an open and a courageous conversation about these issues and in particular about how we go about aligning tech-based innovation with the ultimate well-being of our society and how we might sometimes get pulled off track it will come as no surprise to many in our audience that these are particularly hot particularly focal issues in the domain of artificial intelligence which many refer to as ai i think it's important to recognize that we are in a difficult position as everyday users and consumers of ai it's often easier to kind of start to use it start to rely on it just become kind of enmeshed in the world that is created by it without really understanding a lot of the details about how it works which puts everyday users in a difficult position when it comes to informed decision making about how well ai is actually aligning with their individual goals and with underlying sort of societal goals so when we're you know when we're sort of browsing the the web and getting product recommendations or um when we're trying to buy a ticket from virtual customer service agents when we're getting driving directions or fully automated driving assistance from our cars it's difficult to know whether or not the reference information and the reasoning that's being used to sort of to guide us is actually aligned with our preferences with our underlying individual and societal uh objectives and so that can be a challenge it also it sort of adds the difficulty that in in contrast to situations where we're dealing with guidance from real world human beings with ai systems we often cannot just turn to the system and tell us you know ask it to tell us where it's coming from and kind of why it's doing what it's doing and that kind of adds to some of the mistakes so it means that producers of ai have to do a lot to sort of earn trust and to build confidence among consumers they have to uphold the responsibility to do good with the technology while making sure that they're still pushing the boundaries and they're still exploring and i'm doing the work of innovation which is fundamentally unpredictable you know you're going to be developing these technologies where the capabilities and the impacts can never be fully sort of mapped out uh in advance and we know that that the the firms that are doing this are going to be up against um sort of a little bit of skepticism right so we know according to 2020 financial times data uh that it's about a third of of consumers only these days who are comfortable interacting with businesses that interact with them through ai we know that fully 53 of consumers feel that ai systems will always be sort of making decisions and providing guidance bearing the uh the biases of their original creators we know that only about 12 percent of consumers in those data feel that ai systems can reliably distinguish good from people right and so when those kinds of when those are the sorts of feelings that your consumers have to be a producer to be a developer of ai i think is a uh an enormous challenge and an enormous responsibility although it's also notable that that there's quite a lot of enthusiasm for ai we know that just if you look at everyday consumer behavior we clearly enjoy the fruits of of all of this technology we are you know using the recommendation engines for everything from you know where to eat uh what restaurants to go to to sort of what romantic partner is to try out you know we're definitely enjoying the efficiencies that we're getting from ai-driven search and navigation uh we're certainly appreciating the efficiency and the effectiveness of ai driven healthcare systems and all of that so i think the imperative for leaders for firms in this space is to try to figure out how to kind of maximize and sustain the advantages of artificial intelligence while making sure to identify and minimize any of the aspects of the systems that can do us harm okay so that's going to be the sort of context in which we set this conversation and to help us reflect on these important issues and considerations i could not be more delighted than to have our special guest for today dr john kelly uh who is uh a renowned leader in the tech space a driver of innovation in almost any area you can think of when it comes to information technology recently retired as executive vice president of cognitive solutions and ibm research after over four decades of leadership there perhaps most widely known as the father of watson the famed ai driven i think two-time jeopardy champion that ibm produced dr kelly actually put ibm at the forefront of innovation uh in everything from semiconductors to supercomputers to many other technologies he's actually done so well for ibm that they have been able to maintain their status as a u.s patent leader for the last 28 years running straight you can also see his impact in in terms of driving transformative collaborations and these ones can span the globe so he created a network of ibm labs that engaged 3 000 scientists and technical workers across i believe 12 labs and 10 countries and more recently in a really uniquely boundary-spanning collaboration brought ibm together with microsoft and the roman catholic church to become initial signatories to the rome call for ai ethics unsurprisingly uh dr kelly has received numerous honors and awards including three honorary doctorates to join his initial phd in mechanical engineering from ren sailor polytechnic institute he also won that institution's lifetime achievement award he's also gotten the national academy of engineering's arthur m bush award for driving excellence through corporate government and university collaboration he's an excellent guest to bring to to help us think through these issues i want to welcome him john thank you for joining us good morning heather great to be with you all thank you i want to get started john by thinking about your position seeing multiple transformative technologies uh sort of rise and and um and and become mature over the course of your career in the context of that experience i want to know if you think about the ethical conversation that's happening now with artificial intelligence as distinct from the ethical conversations that typically arise and have arisen with those new technologies or is there something meaningfully different about what you're hearing today sure sure well heather first my thanks to you and the whole team at ucla and uh chris lowe who is very helpful in uh reaching out and bringing me in so thank you all for this opportunity um as you'll see this is a topic that i'm very very passionate about so so to your question um there's there's many similarities with um the ethical implications of ai uh to what i've seen in other technologies but there's a big difference um so when you think back you know all new technologies raised ethical questions going back to the steam engine atomic energy the internet all of those um which were technologies that were advancing at sort of exponential rates caused people to really pause and think about well what what are the the good and the bad uses of this technology what are the ethical implications what i think is different about ai is that it so much gets at who we are as humans because it's it's coexisting augmenting our cognitive capabilities and whenever you're doing that i think we're into a different level of ethical issues and it sort of relates to who we are as humans and you know i'll never forget a couple of conversations or a couple of conversations and instances uh you know when you you reference watson winning in jeopardy and and of course it uh it beat two champions but ken jennings was was the uh the all-time winner i i later saw ken on a uh a ted talk where he sat in front of an audience you know when i when i lost that match i've gone through a process where i really don't know who i am because that game and my championship was who i am and and now a computer took that another conversation i found very interesting was with gary kasparov um before we did watson heather we built a machine uh 10 15 years before that which was became the the grand master of chess and beat gary kasparov a years and years later i had lunch with gary and a couple of other people and he was obviously very upset when a machine called deep blue at the time beat him um but someone at the lunch asked him said gary would you rather have been beaten by a machine or a person and he paused and then he he he got very uh animated and he said no you don't understand it doesn't matter if it was a machine or a person i never lost a chess match i never lost and now i lost so those two things just strike me that with ai we're getting so close to who we are as humans and how we identify ourselves but i think it raises it to a whole new level of uh ethical requirements heather yeah i think that's an excellent point and it it drives it something that i think is is probably part of every innovative push and effort but maybe has particular significance in this domain which is you know when you're driving that innovation you have to balance two imperatives you have to balance the imperative to um to disrupt and to sort of teach everybody what we didn't know before to surprise everyone to see what role these technologies can play in our lives recognizing we can't always foretell that that's the one imperative on that side but then there's the imperative obviously to do no harm and when we're talking about technologies that get so close to who we are as human beings understanding the various kinds of harm that people can feel the visible and the invisible the internal the emotional harm those kinds of of things get to be quite complex so i i wonder if you could reflect over the course of your career on how you have balanced those two different imperatives and how you would recommend that leaders in general in this space think about balancing that those two sure sure you know it's um there's um you know it's the old there's a ditch on both sides of the road you can you can almost become paralyzed and say well i shouldn't innovate i shouldn't invent this because someone might take it and do something wrong with it on the flip side and this is always how i've thought about it um if it can be done it will be done by someone and i think it's incumbent upon us then to have the the most ethical leaders the most ethical people doing it and to recognize the ethical implications and have open discussions and set a set of principles in terms of how to guide one's decisions in this area and i will probably talk more about you know how you set those principles and how you think about them but you know think about technology like ai it's advancing so fast just in its raw capability you mentioned lots of the applications it is everywhere and it gets smarter every time it's used and it never forgets so it's not like when you have a technology like that it's not like you can sit down and write you know five rules and it covers every case of fai so um what i have always found the best way is to drop back to whatever set of principles that we can use in guiding every decision we make every day in both inventing the technology and using the technology it's a really interesting point that these principles can be used as a sort of beacon towards which we we aim when we're developing the technologies i want to ask you in some sense to think about where those principles come from so as you know the the eu recently unveiled a draft set of regulations sort of um top-down principles in a sense aimed at regulating commercial and governmental use of artificial intelligence across a variety of applications from self-driving cars to hiring processes and banning some things altogether like automatic facial recognition in public spaces do you think of that sort of regulation as effective or feasible in terms of really being the driver of ethics and ai well here's here's how uh we approached it at ibm you know in the early days of watson and ai uh we could see clearly the ethical uh implications of the technology and and you know in ibm we have sort of a history of uh not only do we invent it but you mention all the patents but we also believe it's part of our job to introduce it uh ethically and so um i'm actually sitting today in the conference room in ibm research the watson research center where i first saw watson in action and and sitting in this chair and i remember a chill went up my spine heather and i said oh my god this is going to change the world it's going to change the world and that was about a year and a half or so before every the rest of the world saw watson so i started to think about and leaders in ibm started thinking well how are we going to deal with this and so we decided to go principles-based um and by principles i mean for instance just a few of the ones we derived which is it it it takes data to train these ai engines these ai computers and the first principle we set was your data is your data and we have no right to take it use it without your permission or resell it because your data is now part of who you are so your data is your data the second kind of a principle became if we use ai in a product we'll tell you that we're using it so it's not some hidden thing in the background that you're not aware of we'll tell you uh if and when we're using it and a third example would be since ai is a machine that's trained by humans and data um another principle we established is we'll tell you who trained our ai machine so for instance if it's in healthcare we'll tell you that memorial sloan-kettering doctors trained it and not someplace else so a lot of these are are around transparency so as we advance those principles though it became obvious to us uh a few years ago that they were wonderful and they were helping the 400 000 ibm employees and developers of of applications think it through but we came to realize that we also needed regulation but we were very specific we needed precision regulation so because each use case of ai is dramatically different and it's very hard to just blanket regulate so we became very vocal proponents of precision regulation around specific uses of facial recognition or voice recognition or other kinds of ai so we believe that the best balance is principles based and then very precise regulation versus blanket regulation which might actually harm the advance of the technology it's a wonderful way to strike the balance and to make sure that you're incorporating individual dignity and some um some recognition of the importance of of external stakeholders being part of the conversation um and and that brings me to a question about where you think the principles for any given leader within a firm should come from who's whose principles do they use if you're you know if you're the evp uh or if you're a developer on the ground whose principles are you referencing and how do you ensure that those principles are are interpreted consistently throughout your organization so it's not like you have different sets of principles being used by the different people who touch the technology yeah well you know in the end heather um the principles i think are based on an individual or a group or a company's sort of moral values that's the sort of the fundamental base and so you you always want people that set the principles you know to to be the most ethical and and moral um particularly when you're dealing with these bleeding edge technologies that are going to change the world and are changing the world i will admit though it gets complex because you know as you know as soon as you start to talk about morals you're also into cultures and different cultures have different moral standards and values so um in a company like ibm you know we're in literally every country in the world um you know in large numbers and so we really think carefully will this principle translate across cultures but we also believe that while you know we we might be different in the us versus you know europe versus china versus india the company of ibm has a constant moral standard that the company is built on and so we try to make those principles and values so bedrock that they do translate across and then we as a group of leaders you know get together we've we had i can't tell you how many rounds of discussions on those principles and you know we reached the conclusion on them and you know i will admit some wanted to become very specific but i reminded them that you know we're in the early innings here of ai and so they have to be durable and last over time because you never want to put out a principle and have to reel it back in later so it's dynamic but it's really bedrock yeah i like that that notion of you know any individual leader needing to think about the extent to which the principles that they're using really reflect the principles of the organization and to be in conversation with the rest of the organization to make sure that they are in fact in touch with that and and that the articulation of those values and those morals is is flexible enough to go across the different cultures to be understood well across the different cultures while still being bedrock and still being firm and that makes me just want to check in on an implication of this to your point you know when you're talking about morals and and ethics it's very difficult to say okay we we said we were committed to this today and then like next year not so much but to the earlier part of our conversation about always being at the forefront always kind of pushing into a space that can't be completely predicted how do you think about finding out that you were wrong about something as a as an as an individual developer as a firm what's the process for learning from experience and updating the perhaps the the way in which we serve our ethical principles yeah so you know while it's bedrock for us um we also learn as the technology develops um and because our we do so much very advanced research our our high beam headlights into where the technology is going to go um for at least the next decade or two is is usually pretty good and that allows us to um you know establish principles that are enduring that said we do learn uh from it and you know as an example the topic of you know facial recognition as you mentioned as more and more data became available on biases that were being developed in those ai machines we went into high gear we really started we changed our ai engines to remove as much of the bias as we possibly could and and then ultimately we said that look um we're we're not going to continue to sell products based on facial recognition because as good as our principles are it's it's almost impossible to completely remove bias because remember the machine learns based on the data and whatever sample data you give it is what it learns not unlike humans by the way that's how we learn and so we just we just realized that that in that particular situation the risk reward for society wasn't worth it and we just we changed yeah oh that's a great a great story i think it's it's not told enough the the way in which learning experiences in this space can sometimes teach you that you don't want to go any further in this domain that you actually want to pivot and focus elsewhere so thank you for sharing that um i'd like to give our audience a chance to hear a little bit about one of the more recent and really sort of extraordinary collaborations that you have put together to help further drive innovation and to center ethics in the development of artificial intelligence i'm talking here about the room call for ai ethics and the way you brought ibm and microsoft together with the roman catholic church to become the initial signatories of that of that document can you tell us a little bit about what it is and your involvement in it and what led you to think that ibm should become one of these initial signatories sure so um it's sort of an amazing story um so i'll keep it brief but as we were wrestling with you know all of the moral ethical implications of ai just a few years ago amazingly the pope was thinking about the same thing and so he turned to his pontifical academy which is think of it as the pope's think tank and ask them to really think about and study the implications of artificial intelligence he as i understand did not quite understand it but he understood the potential and interestingly enough he was interested in making sure that it was used ethically but available globally because he saw it as a tremendous tool to advance humanity but it could also be another have and have not uh in the world and he was very concerned about a new sort of ai divide emerging so he charged his pontifical academy um they came in uh the united states and and visited ourselves for a couple of days as well as microsoft and through those discussions we realized that we were tackling and thinking about the same challenges and problems through different lenses but ultimately we were worried about exactly the same things and so we were uh we and microsoft were invited to work with the pontifical academy to develop a set of principles um that the catholic church could also endorse and uh that then became the rome call in and around ai ethics which um myself and brad smith from microsoft went to rome and signed just before covid completely broke out by the way and put us all back on the ground but it was a uh an amazing effort and it's um who would have believed that ibm microsoft much less the catholic church would come together and work on something but it i think that in itself tells you how important this topic is that's i yeah it's a really extraordinary step forward and it's funny i think it was microsoft's brad smith that said that that people might think of ibm microsoft uh and the roman catholic church is sort of strange bedfellows in an endeavor like this together um and that that just sort of makes me wonder what you think of the process of that and how on such a complex and multifaceted issue three such you know powerful and independent entities with very strong sorts of principles of their own walking into uh into that space how do you collaborate on something like this how do you come in the end to a document that you can all agree on coming from different perspectives especially with ibm and microsoft in some some spaces being competitive i mean how do you balance all of that yeah it's uh that's a great question um i'll tell you what we did and and how it worked and it's a it's a um it's a process that i've used in other international you know sort of partnerships um we didn't you know in the first couple of days of meetings we didn't sit down and try to write the rules or the principles um we didn't you know well here's 20 rules let's vote on 10 and that's what we'll sign um we started with a an understanding of each other's values and uh morals i guess and you know as an example the pontifical academy was as interested in learning about how ibm thinks about the future of computing like quantum computing and what what is that going to mean to society as they were artificial intelligence so we went off into these other you know dimensions and issues and we took them in different areas and what we we did is we basically established a trust that we were all seeking the same thing we all had the same fundamental value system and therefore it is possible that we could agree on a set of principles and that's exactly what happened so after that couple of days together we we all realized that we're pretty much cut from the same cloth from a value standpoint and ethics standpoint and then our teams could get to work on what specifically were the principles going to be in the wrong call so we didn't we didn't gun jump you know to try to get to an answer which is you know in tech and for type a's we generally try to do that it would set back and as as humans let's make sure we're on the same page before we get started i appreciate your underscoring that i think that's something that we try actually to bring into a lot of our our classrooms because we do have a lot of type a personalities a lot of students want to jump in and solve problems and and there's a tendency when coming into into a room with other people to start by figuring out well what do you want to do and what do i want to do and you jump right into some kind of negotiation of your positions on the issue uh and what you're describing sounds like a really different and also just a richer and a more robust way to build an actual relationship with people which is first of just find out what motivates you what gets you into the room uh to begin with what do you care about in the in the broader scheme of things and then once you understand the extent to which you are aligned on those broader principles you can come together to talk about specific things that you might collaborate on together yeah you know i think if we had tried to just jump to the answer we would have made we would have you know going to the white board and said well here's a list of good good uses and bad uses do we all agree um for instance we would have completely missed the pope's concern of equity of use and and we had a long discussion about that um if he was very concerned that you know areas of south america would get access to it the poor would get access to it that particularly when it reached applications like healthcare or education um it would reach the young as well as the elderly and that it wouldn't cause a further divide in any dimension of humanity and that's that's a dimension we would have completely lost if we had just gone to the white board and started listing good and bad that's a great point and i wonder as you think about the future of the rom call what are the kinds of things that you think will draw other firms other religious institutions other governmental entities into that collaboration as you think about its expansion what do you think is going to pull people in and are there any barriers that you think might might create bumps in the road yeah there's there's been tremendous interest uh in follow-up we obviously we ran head-on into the covet thing which uh slowed us down a bit but there's um we've had meetings with many of the other great religions of the world that are very interested in this many uh big corporations um are studying it carefully because they've many corporations have been busy writing rules and they realize that the technology again is changing so fast that they can't keep up writing the rules so let's go back to the basics and they're looking at this as sort of a foundational um document and set of thoughts um so our intent is to follow up with an additional with additional signatories but also it's viewed you know as a living document we're going to learn a lot and i'm sure we'll be back together as a team you know looking at well what's what's transpired in the last couple of years and what do we need to update in those in that document in the wrong call this seems like a great space to acknowledge the enormity of the problem that faces every one of these these institutions and um and to highlight the power of collaboration is something that can actually help everybody to get a little bit more of a handle on it i'm thinking about that i want to sort of help us transition a little bit more to thinking about your particular career journey and i know that collaboration has been something really at the center of a lot of what you have done could you speak a little bit about the role of collaboration for you as a kind of leadership tool alongside obviously all of the elements of leadership that have to do with making sure that your firm is competitive how do you think about the role of collaboration as part of that sure another fantastic question so you know i've always felt heather that you know as big as ibm is and as much as we spend in r d and all the patents we generate we don't have a corner on the market of smart people and ideas uh much less the means to bring them to market and so i've always used collaboration even with competitors as a way of advancing our technology and our business we did it you know at big scale and semiconductors with other companies with the samsungs and toshiba's of the world companies that in some cases competed with us a second dimension to collaboration though is around open innovation and so we've always been and i've been a huge supporter of open source uh software uh where it's it's not developed strictly in one of our labs some place in the world uh it's it's the community uh we contribute to the community and we take the open source software and then we build products above it and below it but the nugget um of say linux as an example is an open source community and we um we nurture that comm that community we help we we put code into it but we do get the benefits back uh in that regard so another dimension of this collaboration are these open communities and you know we see it as a force multiplier for our own r d um as as well as it's just a fantastic way to bring new technology into the company wonderful i want to zoom out a little bit and take a look at your incredibly long and impressive career i mean it's so rare these days to find people who have the benefit of long tenure in an organization and you have four decades at the forefront of innovation at this incredible pioneering firm i want to ask you to sort of reflect on the balance that you struck during that four decade long sort of tenure making incredible transformative change on the technological side we talked about the semiconductors and the super computers and ai and all of that but also driving incredible uh transformative change on the business side of the house in terms of partnerships in intellectual property and security and privacy can you tell our audience a little bit about how you transitioned your stem background and your technical work into roles that unite technology and business functions sure so when i uh i joined ibm i i literally heather defended my phd thesis on a friday and started at ibm monday back in 1980 but when i joined uh to be honest and transparent all i wanted to do was technology it was just just put john in an r d lab and and i would be happy um but i i quickly learned that and this is part of why i came to industry versus academia at the time i quickly learned that to get things done and to have an impact at any kind of scale i had to i had to pull teams together and i had to understand the business side and that is what caused me to finally understand what i was best at but also um what was most effective for the company and i always sort of describe myself as being on the diagonal and what i mean by that is if you if you view sort of business this way and technical this way i'm like on this diagonal and i've wandered off the diagonal at different points in my career pure business pure technology but but by and large for four decades i stayed close to that diagonal and i'm just very fortunate that it's what i enjoy most if i'm doing just technical or just business or any extended period i'm not very happy but it turns out that i have found that a company like ibm and many of the other tech firms uh that you not only not only am i happier but you can get a lot more done and have a lot more impact on the world i mean think about if we hadn't brought ai to market uh we wouldn't have changed the world um you know we're now bringing quantum computing to market we we could have left it in our labs and you know advanced qubits and things but so what so we've got to bring it we've got to bring it out to the world and that that's where business comes in yeah that's a great point and it brings to mind you know the sort of stereotype of those who have great technological skill a great sort of focus in this area the sort of stereotype of individuals in this area is lone wolves who want to get all of their work done independently who want to make their career success on the basis of their individual brilliance and i think that the existence of that stereotype can often overshadow the importance of the ability to work within teams to pull together teams to build relationships that help to to maximize the value of the individual technical skills that people have i want to ask you to reflect on on those kinds of relationships maybe in particular mentorship relationships which is something we talk a lot about here is something important for our students to cultivate can you tell us about the role of mentorship in in driving your own career success sure so sort of segwaying from the last discussion i i always i always believed heather that it wasn't about just inventing a technology or leaving a piece of technology out there that you know as a leader um one of your most if not the most important legacy is the team you leave that the team you build and what that team then goes on to do and you know i i'm i'm very proud whether it's the ai team or semiconductor teams or super computing teams those teams go on and on and on to do bigger and better things and that i think is a contribution that goes way beyond you know uh the thing itself so on on mentorship um i again i've been extremely fortunate and blessed um i've had mentors both in ibm and outside but probably not traditional mentors not mentors you sit down with and have a chat and they they sort of coach you and well john you got to do more of this or more of that or try that kind of job i had very little of that what i did have though is a set of mentors and network that would uh challenge me and give me experiences so throw me in to something that was much bigger and much more difficult than i thought i could possibly deal with or handle and it's sort of sink or swim throw him in the deep end of the pool and see if he comes out but but what you learned from that is incredible you really learn how to get things done how to work with people but you become more confident in yourself um you know okay throw me in a deeper pool and i i'll find a way um through it and so the best mentors i've had are the ones that they don't sit down with me you know one hour a month or whatever they just they're always around and they're like okay john go go do this see if you can take that on um and it could be technology it could be business it could be a people issue a leadership issue a government issue but just you know go go try this one and through those experiences then you just build up this incredible sort of well-rounded capability it's really interesting i think a lot of our students um tend to think of mentorship initially as something that's meant to make your life easier like you go find somebody who's done it before they can tell you how to do it and then you're um then you're on easy street and what you're saying is in some sense the opposite find find people who are going to make your life harder absolutely dude for the students uh i mean if you enjoy sitting with people who don't challenge you that's okay but find people who will you know throw you in a deep end of the pool and the other thing i would say is um don't be afraid to take on really tough things um you know i'll admit that at first i was afraid to do it you know could i really do it um and then i started to think well gee people will start to associate me with okay there's a problem john's always around uh so you know which is chicken and which is egg um but i i found that the biggest challenges whether it's in technology or business attracts the best people in the best minds and so i always sought to go to where the problems were because i knew i'd be rubbing shoulders with brilliant people and maybe some of that would sink in uh to me and so run to the problems and if you can't run there have a mentor that will throw you at the problems and amazing things will happen i love that i love that um i am going to see if we can use the last little bit of time just to make sure we get our audience questions in give me a second i'm just going to try to pull up those so our first question is that dr kelly mentioned that ibm has a solid set of guiding principles how do we ensure that companies without principles are held to the same standards well that's that's an area where uh precision regulation has to come in and and that is uh that's why we realized that and crossed that bridge because you know not everybody and every company will have the same set of ethics and so we do need the precision regulation to protect against the corner cases and the more most harmful cases um to society a second thing uh that can be done and and this is to help people abide by the principles you can actually build into the technology things that will monitor and compare behavior to your principle and so as an example we built a set of software that all of our developers use in ibm which is constantly monitoring their code to see whether the code is structured in a way to develop a bias or whether the data they use to train it is developing a bias think of it as sort of a bias meter and so as you're writing code it's going okay you're getting much more biased you better come back and what we find is that if most people uh they're they're not conscious that they're doing it but if you show them what they're doing most people will mid course correct if they understand what they're doing so i think the precision regulation but also giving people the tools because most by far most people are ethical and if they realize what they're doing they'll react accordingly yeah and it highlights the opportunity to put yourself in an environment where you are getting kind of early warning signals about uh the alignment of your part your principles and your uh and your behaviors rather than having to only be subject to something like the regulations from external entities that's right another example would be uh you know one of our principles we don't we don't leave holes in our code you know what are called back doors in software that somebody or some government or whatever it can come in through um and so that's a that's a fundamental principle and so our people that write our software and our code are very conscious not to be sloppy to leave a door slightly open if you will or a window slightly open make sure take the time to make sure that every door is closed and locked in every window so that the code is protected from from people who might want to do harm to it that's excellent there's a question about whether or not you see an arms race in artificial intelligence and how that might be influencing the moral foundations of ai well um unfortunately um like most technologies these are all multiple use technologies and so um there is a bit of an arms race it's both from an economic standpoint but also military i mean most of the major military powers of the world including the u.s have openly talked about leveraging ai in defense systems and offensive systems so in that sense it is a bit of an arms race and this is where again i think we need like we did with the internet and cyber security and we still need we really need more international discussion around these topics what is acceptable use of the technology i'll give you a very specific example one of our principles is that we will always in potentially life-threatening conditions we will always make sure there's a human in the loop we will not allow an i ibm ai system to take data use ai and then cause something to happen without having a human in that loop either on the intake or the decision process and we think that's really a fundamental important principle um in a in an arms race if you will so again i think if we get back to the principles we have more international discussion um that would be uh wonderful and i go back to you know the rom call you know if if microsoft ibm and you know and the catholic church can come together then i think anyone can come together on the right principles i think it's certainly encouraging and i actually i want to jump to a question that also sort of helps us to dig into the question of how humans can be in the loop or how humans and and ai can complement each other and drive one another to better uh to better outcomes the question is i wanted to know if there are any advances in using ai to help humans learn differently so that humans can compete with ai at least in decision making yes so um i would i would strike the last part of that question um which is to compete with and i i can't emphasize enough that this is human plus machine and there have been countless studies in all sorts of domains heather that um hit basically in a challenge a human plus a machine against either a machine or a human and in every case in every field human plus machine always win always can beat one or the other and that that has stood true since uh since the early days of watson um to the point now where uh gary kasparov when he plays chess as an example um wants his matches to be he wants a machine with him i wouldn't want to play chess against gary when a machine that would be an impossible thing to win but um think of it as human plus machine as opposed to can we cause one to be able to beat the other um you know i have to i have to admit that um i don't know if it was right for me to to use whether it was deep blue or watson in demonstrations of human versus machine because in a sense it it drove that kind of thinking which causes this kind of question if i had it to do over again i think i would do a human plus a machine versus a machine and a human to show that power and i guess i would just end this this particular question with that puts so much burden on the human machine interface and optimizing that and we now have so much work going into how do humans best interact with machines and how do machines now read and interact with us the natural language processing of these machines is actually better than humans yeah but how does the machine understand your tone your position your face your reaction and how does it sort of impedance match as an engineer with the human and that whole field is is is just open for great research i really love that point and i'll i'll finish this up there just because i think that notion of of the addition the human plus machine these the idea that we're talking about some things that come together in common good to make everything better on all sides i think that's been something that has come up multiple times throughout our conversation um it helps to explain why i think the sort of the principles based uh approach and the principles that you were able to come together around when you came together with microsoft and the roman catholic church sort of getting down to that root what what brings us all here what is that sort of common benefit for humanity that we're all striving towards if we can get there then i think the the specific differences we might have or the discussions that we need to work through in order to drive uh ai and technology forward become much much more manageable and much more fruitful so with that i just i want to ask you give you just one last chance to say you know is there anything you want to add to share with our community any upgrades on what i've said well i think i would end with one thought um which is these machines are just a reflection of who we are as humans you know i was once asked uh john will watson does watson have a soul will watson choose a religion you know etc etc and i i reminded the person that this machine was built by humans it's trained by humans it's programmed by humans um it's corrected by humans so whatever set of ethics it develops or whatever religion is totally up to us and we need to be very conscious of what we're doing with these machines because in a sense when we look at these ai machines we're looking in the mirror we're really looking in the mirror thank you so much john kelly for joining us it's an incredible pleasure to have been able to spend this time with you thank you to our audience and thank you to terry and the easton center for putting this on have a great day thank you heather thank you at all bye-bye excellent let me just add my own uh thanks to john and to uh heather what a great way to start and i you know i want to share a couple of my so what's i don't want to wait to the very end because i think these are cumulative in nature and matter a lot and i'm going to start out when john talked about personalization and what created problems people feel like you know who you are is being exposed maybe not in a good way well if you put that aside for a second and think about what technology does it allows for personalization you think about recommendation engines it tries to get a better understanding in areas of media to know what you like to make relevant recommendations if you look at health care so this dichotomy of personalization which is a core benefit of ai is also something that makes people feel not so good and so you better design it well to get the real benefit if personalization was a nothing issue then you would say forget about all of it but it's it's one of the big benefits second comment that john made that i thought was a very important call to action if you don't develop the innovation somebody else is going to do it so to think that the answer is let me stop all my development let me not develop products and services that are technology based is not a good answer because somebody else is probably going to develop it and society is going to have the same issue and companies are going to have the same issue third thing is what are the imperatives of leaders here and i always find this interesting so we get past what could be a theoretical discussion john talked about at ibm a variety of things that they use as their tools to make sure the technology is wisely managed he talked about transparency principles he talked about precision regulation and he talked about bias meters all his examples and they're heavily oriented around what you as a leader and you as a company own as opposed to what i call punting stuff you know bring regulation on and it's like what does somebody exactly mean when you say bring regulation on so i thought those points were excellent a couple of last things this human plus machine is better than human or just machine i have to admit when i cover in my class we talk about the growing capabilities of machines and they will eclipse humans at some point with lots of data john's making a very important point when you're talking about pushback concern etc even it can be on a basic level about adoption of ai if you're pitting the machine against a human you're going to get a real sense of pushback on that on that opportunity so think about human plus machine as where is where the real love benefits are final comment he made again a real commentary about individuals in society the machines are just a reflection of who we are that can either be a good thing or a bad thing and i think what he's saying is be a good person develop the right intentions etc and you'll get good products don't do that and you're going to have a big problem on your hands and i think he's indirectly calling out there been some companies that have not managed this stuff very well shame on them um let me just thank heather i just want to thank you again for eliciting a great great conversation that is truly interdisciplinary there's a living example of kind of transcending boundaries as you think about business and leadership so huge thank you well done
2021-11-25 12:09
