View from the Top: Thomas Siebel, C3.ai

View from the Top: Thomas Siebel, C3.ai


– Alright, good morning everyone. Why don’t we get started? Oh actually, good afternoon, everyone. Welcome. I’m Tsu-Jae King Liu the Dean
of the College of Engineering, and it’s my pleasure and
honor to kick off this first View From the Top event
of the fall 2019 semester. This speaker series is
designed to bring leading leaders in business and
technology to speak to our College of Engineering community, and today’s special
guest exemplifies this. He is a visionary Silicon
Valley entrepreneur, Tom Siebel. But before continuing,
I’d like to acknowledge the co-host of today’s
event, that’s Tau Beta Pi, the engineering honors society. Let’s give them a round of
applause for helping us. (audience applauds) I’d also like to extend a warm
welcome to all the members of the Dean’s Society and
also everybody who’s watching and joining us online. Before going on, I’d also
like to remind you of a couple of events that
are happening next week, so on Wednesday at 2:00
pm, we’re going to have a College of Engineering
community event to celebrate the 10th anniversary of the
Engineering Student Services. Oh, October 8th, I’m sorry. Okay, we just postponed
it because ESS advisors are not available next week. Alright. So, on October 8th, we’re
going to have a celebration of the Engineering Student Services, which really is designed
to help to enhance our students’ learning
experience, so 10th anniversary and a new welcome center for the college, so mark your calendars for October 8th. Next week, the next speaker in this series is Chandrika Tandon. She is a businesswoman and a
Grammy award winning artist and a humanitarian, so
she’ll be joining us on this stage next Friday. So please look for the
announcements for these events in your email. Alright, so coming to today’s event. I think we all are aware
that advancements in information technology over
the last decade, few decades, have really lead to faster and
most cost effective devices for communicating and processing
and sensing information in our world, and this has, in turn, transformed the way we
live, work and play. So advancements in information
technology have led to cloud computing, the Internet of Things, which in turn have enabled advancements in artificial intelligence
and machine learning and also big data and data science. So the confluence of
these four technologies is going to affect a digital
transformation of businesses and governments, which is going to have profound implications on our society. Now, you might know the
Berkeley Engineering has been enabling, an enabler in
innovating new technologies that have enabled the information
technologies revolution and we expect that we’re
gonna be at forefront of the A.I. or digital
transformation revolution. But also, this revolution really
has some serious potential societal and ethical implications,
and as part of the larger Berkeley campus community,
we should take advantage of the breadth of excellence on our campus so that we can be leaders in
addressing all the challenges that digital transformation will bring, along with the benefits. So this is why I’m especially
excited today to have our speakers, our guest
speaker talk about his vision and his advice for leading
the digital transformation of our society. And at this point I’d like to
invite Professor Costas Spanos to come up here and set the
stage for our speakers today. Costas. – Thank you, Tsu-Jae, and good afternoon. Very nice card. So welcome all to Citizen
of Banatao Institute, that is co-hosting this event alongside with the College of Engineering. And today, we’re extremely glad and proud to introduce our speaker, Mr. Tom Siebel. Tom is the Founder and Chief
Executive Officer of C3.ai, a computer software company that provides platform as a service and
software as a service applications for enterprise scale big
data, predictive analytics, A.I. and IoT solutions. Tom was the Founder and
Chief Executive Officer of Siebel Systems, one of the world’s leading software companies. It was founded in 1993 and
Siebel Systems pioneered the CRM, customer relations
management software, and it became the global leader, with more than 8,000 employees
worldwide in 32 countries and over 4,500 customers and annual revenue in excess
of $2 billion in seven years. Siebel Systems merged
with Oracle Corporation in January of 2006. Mr. Siebel is the Chairman of the Thomas and Stacy Siebel
Foundation and serves on the College of Engineering Boards of the University of Illinois and the University of California Berkeley. He’s been a very good
friend of our programs here and we’re very grateful for that. He’s a Director of the Hoover Institution at Stanford University and is member of the American Academy of Arts and Sciences. Tom is a graduate of the University of Illinois
at Urbana-Champaign, where he received a Bachelors
of Arts in History, an MBA, and a Masters of Science
in Computer Science. He’s the author of four books,
including most recently, “Digital Transformation:
Survive and Thrive in an Era of Mass Extinction.” Today’s subject is gonna be,
indeed, digital transformation in a conversation with our
distinguished colleague, Professor Shankar Sastry. I would like to welcome
you both to the stage. (audience applauds) – Thank you, Tom, thank you for being here and thank you all, full house here, so. Delighted to see our campus community. I’m gonna plunge right into it. Just so you know, we’ll
have, I have a few questions. We’ll get going with this
conversation, but we wanna throw it open to questions
before not too long. This, by the way, is the book. I hope you’ve already read it and. I got to re-read it in preparing for this. It’s just a, I would
say I really enjoyed it. Ah, here it is, okay. I enjoyed re-reading it. So Tom, let me start off by
asking you about Daniel Bell and his 1973 book, “The
Coming Post-Industrial Age.” As you say in the preface of
your book you didn’t miss much. But Bell’s book is
really not standard fare. How did you come upon this book? – I spent a lot of time at
the University of Illinois in Urbana, and my first degree
was in history of science, and then I went back to learn
the languages of commerce so I could play the game of
business at a higher level, and so I went to the
Graduate School of Business to learn the languages
of finance and marketing and accounting and what have
you, so I could play that game, and I, the course that I really enjoyed was a course in operations
research which brought me down into the computer lab and back then, we were computing on these
massive CDC cyber super computers and coding in fortran on key
punch machines with large decks of hollerith cards that
we’d take in every evening to have read and we’d
get to the third card and there’d be a typo, so
you’d have to go do it again. Anyhow. In the course of, you know, I spent a lot of time in the computer lab and got involved with the discrete-event simulation
and whatnot, and one day I was roaming through the
bookstore in the Illini Union, and stumbled upon book
published by Daniel Bell. It had been published in
1973 and it was called “The Coming Post-Industrial Society.” Now let’s think about this. 1973, this would be the
era of the Ford Mustang, this is the Interstate
Highway Act, Detroit is king, we are in the business of
manufacturing goods as an economy. And so Daniel Bell, it turns
out, the title of his book is “The Coming Post-Industrial Society,” and he was a Marxist sociologist
from Harvard and really quite influential in the
’70s in the United States, one of the most influential
thinkers, and his thesis in this book is the global
economy is about to go through a restructuring in the following decades on the order of the Industrial Revolution, to what he calls the
Post-Industrial society and he also coins the
term the Information Age. I mean, let’s think about this. 1973, I mean people, this
was before the mini computer. This was before the personal computer. If you can imagine this, it’s
before the cell phone, okay? It’s before the fax machine,
I mean, before any of this he puts forth this idea that
we’re gonna see a restructuring of the global economy on the order of the Industrial Revolution
where the scarcest resources are going to change from,
in the Industrial Era, where we have various
forms of capital, okay, to the post-Industrial
Era where the scarcest and most precious resource
becomes information and timeliness and
accuracy of information. And so he kinda foretold this whole story that we’ve been living
in the last, you know, four or five decades and I
found that very inspiring and so I had been taking some classes. If you know the campus at the University Illinois at
Urbana, and some of you do, it’s kind of the demilitarized zone, line of demarkation is
something called Green Street, and north of that is
engineering and south of that is liberal arts and sciences,
and so I had been taking some classes north of Green Street mostly in operations research
and managed to somehow get admitted to the Graduate
School of Engineering, for which I was completely
unqualified, and this motivated me to get
a degree in this field, so I could play the game in this economy that Bell described. And so I did my graduate work
in Relational Database Theory and for the past four decades
have had a place at the table and an opportunity to
play the game and seeing this whole future that
Bell foretold unfold right before our eyes
exactly as he described it. – So, the title of the book says, “Survive and Thrive in an
Era of Mass Extinction,” and in chapters one and
two, you make a case for how the confluence of IoT,
A.I., machine learning and elastic cloud
computing is likely to be really disruptive as
in evolutionary biology and you talk about the
punctuated equilibrium. Tell us your vision of the
scale of this transformation and its punctuated equilibrium. – Okay, so we, this effort
C3.ai, we began in 2009. And as we looked at the world in 2009, it appeared to us that
the next step function in information technology that was likely to change everything included
elastic cloud computing, big data, the Internet of Things and A.I., and now I think that is all,
there really wasn’t much going on in 2009, but today,
it turned out to be true. And so we basically worked on
building a software platform that would allow our
organizations to take advantage of those technologies to
build classes of applications that work to solve problems
that were previously unsolvable. We can talk about those later. But in the course of traveling
around the world and visiting chief executive officers
and boards of directors in Shanghai, Beijing, Rome,
Paris, London, New York, Minneapolis, Chicago and San Francisco, these people with whom I had
done business now for decades, all of a sudden were talking
about this thing called, in every board room
and every CEO’s office, they’re talking about this thing called digital transformation, and candidly I found the idea completely perplexing. I’m thinking, digital
transformation as opposed to what? As opposed to analog transformation? (audience laughs) And the more, and it was clearly something that was very important and it was urgent. It was something that
people needed to do now. But when you poked at it,
your tried to tease apart what they were thinking about,
there was no commonality between their use of the term. Candidly, they had no idea
what they were talking about. And so this goes and gets us from, say, 2009 through about 2014, and then I think I figured out what is happening. There was a, what’s going
on in the world is that in the first two decades of
the 21st century, we’re going through a mass extinction
event in the corporate world, and in fact, in the first
18 years of this century, 52% of the Fortune 500
companies have disappeared. Where is Westinghouse? Where is Kodak? Where is Sears Roebuck? I mean, these companies are
just gone, and it’s almost inconceivable that these
companies can be gone. GE will be the next to disappear. And 52% are gone, so
what we’re dealing with, for CEOs, is an existential event. And then many of you, probably all of you, have read this book called “Sapiens.” It’s a pretty good book, right? And many of you either have
read chapters of or all of, so I read “Sapiens” and this
started me thinking about evolutionary biology and
I had the opportunity to take a class in that area in the past, and so I recalled “The Origin
of Species” by Charles Darwin, and in “The Origin” he
describes this process of natural selection as
the driving force behind the speciation of the planet, right? We all know that story, so we
don’t even want to go into it. And he did this, as I
recall, this was like 1859. Might give you a couple
years one way or the other if I’m wrong, but that’s pretty close. And the problem that Darwin
had is he couldn’t explain the gaps in the fossil records,
and there were huge gaps in the fossil records,
and his explanation was, well, we just haven’t found them yet. And it wasn’t until 1972 that an evolutionary biologist by the name of Stephen J.
Gould at Harvard described. And by the way, Darwin thought
about the process, evolution, as kind of a continual
process that took place relatively gradually over
many, many millennia. It kind of moved constantly
upward to the right, kind of the way we
think about Moore’s Law. Well, Gould came up
with this concept called punctuated equilibrium, and
he said this is not the way that evolution works, and so
I think that the planet Earth has been around for roughly
four and a half billion years. We’ve had life on the
planet for three and a half billion years, and in the
last 440 million years along, which is relatively a blink of an eye in this bigger picture, we’ve had five mass extinction
events on the planet Earth, and it’s through some of
these extinction events, glaciation, tetatonic issues,
the poles switching sides, and during these extinction
events, as many as 96% of the species on
Earth became extinct, and these extinction events would
have long period of stasis, and then an extinction event,
followed by mass speciation of all these new species with new DNA. And so the most recent of these
events is the one with which everyone in this room
is undoubtedly familiar, called the K-T Extinction,
that I happened, I think, about 65 million years
ago when this meteor hit the Yucatan and then this
caused massive climate change as a result of volcanic
eruption and 75% of species on Earth became extinct
65 million years ago. And kind of one of the
predominant species on the planet as of that time were the
dinosaurs, very successful species that had been on the planet for over 150 million
years and it disappeared. And this is followed by mass respeciation, and this story turned
out pretty well for us, because mammals filled that
vacuum that was formerly occupied by the dinosaurs, and
so then we have homo sapiens and so far, this has
worked out pretty well. I know that we’re only gonna
last another 11 or 12 years, but you know. (audience laughs) It’s been a pretty good run so far. Okay now, and so what I, the fundamental thesis of
this book is about drawing parallels between evolutionary biology and what’s going on in the intersection between
technology and homo sapiens. And we see again, these
long periods of stasis and then there’s changes
like the invention of fire, the domestication of agriculture,
the domestication of dogs, the invention of printing
press, movable type, how big was that? This was big event, right,
resulting in the reformation. The steam engine, the jacquard loom, and each of these events
were massively disruptive and in many ways massively positive. At the same time, there were
huge adverse consequences associated with each of them. So let’s think about the press. The movable type. I mean, the implications of
that for the western world were now, you know, the Bible
was no longer hand scribed in one language
interpretable by one group, so we can print the Bible in
French and German and English, and we did, and it resulted in how many hundreds of years of
religious wars that followed after Martin Luther
nailed some declarations to the door of a church,
I think, in Wittenberg? Or the steam engine and the jacquard loom? Well, what followed from that would be the Industrial Revolution, and
most people would argue that the Industrial Revolution
was a pretty good thing. And the people live longer
and they live healthier, they’re better fed, we have civilization in lots of parts of the world. I think most people would argue that it was a pretty good thing. At the same time, you
can draw a direct line from the steam engine
and the jacquard loom to child labor, Communism,
World War I, World War II, I mean, it’s a straight line, okay? And so there were some
pretty significant adverse consequences that resulted
from these technologies, so that’s kind of the big
idea is to draw parallels between evolutionary
biology and what goes on in the intersection of
mankind and technology, and I think we’re seeing
a disruptive event, not less significant
than the jacquard loom, not less significant than the
steam engine or movable type, and I think this is what I think digital transformation is about. This is about the confluence
of elastic cloud computing, big data, the Internet of Things and A.I., and it changes everything. So, if we look at the
information technology industry, when I left college and went to work for a young entrepreneur in Mandell Park by the
name of Larry Ellison. Turned out to be a pretty good idea. The worldwide market for
information technology was over $150 billion. Today it’s $3.5 trillion,
and again, in five years, it will be over $9 trillion,
and this exponential growth is all a function, in my
opinion, of these technologies that I’m talking about. I think 5G might have an impact also. That’s the big idea. – You know, to this audience, of course, they want me to ask all
these technical questions, but what I want to continue
with what you began. So in chapter nine,
moving ahead in the book, you talk about how the
case for digital change is lead from the top by
Isabelle Kocher, at Enel, or Mike Roman at 3M,
Starace Francesco at Enel and Secretary Heather Wilson at U.S. AF. Tell us, what resemblance
do you see in this leadership group that’s
actually making this change to keep them out of the
punctuated equilibrium, the extinction event
that you talk about here? – Okay, so this is my fourth decade in the information technology industry. I’ve seen the transition from
mainframes to mini computers to personal computers,
ethernet, the internet, cloud computing, SaaS computing,
enterprise application software, I have been
there, I have done there, I’ve seen it, and I was
involved in bringing many of these technologies to market. And I would I say, in the last 30 years, these technology decisions
were made in universities and research institutions
and the government and private enterprise by the CIO. The CIO is the person who would
say well, we’re never gonna use a mini computer or we’re
never gonna use ethernet or we’re never gonna use
the personal computer. So you wait for that person to get fired and you come back two years later, okay? And this is how the industry grew. It was always the CIO
who made these decisions. And the CEO was nowhere to be found. If you installed, maybe you
had an enterprise application software system that reached
100,000 people at a place like IBM, then you’d spend
some time with Lou Gerstner or if you did the same
thing at General Motors, or excuse me, General
Electric, you’d get to spend some time with Jack Welch,
but it was definitely the CIO making all these decisions. And the CIO, he or she, would then brief the Chief Executive Officer
once a quarter on what, if anything, was being done. Now, what I found curious as
I began traveling the globe in this most recent effort, say in 2009, is that the Chief Executive
Officer was in every meeting. Okay, and this Chief
Executive Officer was driving every discussion and the CIO
was no place to be found. The CIO was off installing
SalesForce, your single sign on or something or trying to figure
out how to get SAP to work. And the CIO was not in the room. CEO was in the room, mandating
digital transformation. Where did the CEO come from? And so this kind of lead me to this idea of an existential event. And I think you have two
classes of CEOs out there. You have those that are complacently waiting
to get their pensions. They don’t care, okay? All they want is their pensions. They’re not really representing the interests of their shareholders. And then you have these
visionary CEOs who are driving change and taking the
action necessary personally to ensure that their companies have sustainable competitive advantage in the 21st century economy. And these would be people
like by Isabelle Kocher, who’s a physicist and a
great visionary at a company called Engie in France
and my friend Jacques Biot knows her very well, and Engie is a, about a 70 billion year old
integrated energy company. Or Francesco Starace at a
company called Enel in Rome. Enel is the largest power
utility in the free world. They have about 60 million
meters in 40 countries. To put that into perspective,
there are about 100 million meters in the utility
system in the United States, so this is a pretty utility. And by the way, those 100
million meters are serviced by 30 to 150 utilities. Mike Roman, the CEO of 3M, Heather Wilson, the
Secretary of the Air Force, Ryan McCarthy, okay, the
Secretary of the Army. And these people are visionaries,
they are driving change in their organizations, they
are generally aligning themself with a new executive, somebody called the Chief Digital Officer,
who has the license and the authority to get this job done, and they don’t meet
quarterly, they meet daily. And so these are people
who are taking the time to understand the technology,
they are picking up books, they’re reading books and
they’re becoming personally conversant with these
technologies and they are driving change in everything about the
way that they design products and services, manufacture
products, deliver products, service customers, train their
people, compensate people, run the organization. It changes everything about
digital transformation. It is not simply about
installing technology, it’s about changing every manner in which these organizations operate. – Wonderful. You know, there are other
questions related to this change management, which
I wanna come back to, but. I hate to tell you, Berkeley is the guy that did the Yucatan. Luis Alvarez, I think that
was, that work was done here. – That work was done here? – Yeah, yeah, it’s not
a place that’s averse to punctuated equilibrium, the 65
million year old one anyhow. But, okay, okay, no more about it. I want to talk about
you and the book here. (audience laughs) Not about Berkeley is a great place. Okay, in this Renaissance of A.I. chapter, I think you present a very nice perimeter of the multiple cycles of
A.I. and machine learning, which a lot of us in this
audience have suffered through these A.I. renters and so on. And you talk a little bit
about the emergence of A.I. and machine learning,
especially deep learning. Now, some people are
really concerned about sort of the brittleness
of these algorithms. But give us your sense of
how, what’s ready for use and practice and how is it
likely to play going forward? – Well, the concept of
A.I. is not new, right? This came about, what, in the ’50s with Marvin Minsky and these guys at MIT. And again, nothing happened for 70 years and that was called the winter of A.I.. What we were lacking were the underlying enabling technologies. One underlying enabling technology would be the elastic cloud. I mean, with the elastic
cloud, we have all of us, immediately available today,
virtually infinite computing capacity and infinite storage
capacity at virtually no cost. And by the time Saathiya and
Jeff Bezos and Andy Jassy slit one another’s throats,
this stuff is gonna be free. (audience laughs) It will be. So, it’ll essentially
be, it’s approaching. Infinite is a big number. I mean, there’s people in
this room who have done serious computing on 8 bit processors that operated at 300 hz cycles. You’ll remember that. You had to think. And you look at some people
in this room who’ll remember that storage is something
that you would bring in in a forklift into a machine room. Today, with this elastic
cloud, we’re not looking at 8 bit processors at 300 hz
cycles, we could spool up, literally, 10s of thousands
of 64 bit of processors, doing virtual machines doing 64 bit 40 point
operations at 3 gz cycles and store virtually
unlimited amounts of data. So that is what, that was
fundamental to be able to solve these problems because the
data sets are staggering, and the computational
requirements are breathtaking to solve some of these
machine learning problems. So what can we do? So I think about it. Julius Caesar used to talk
about gall divided into three parts, and so I
think A.I. is divided into three parts, and this’ll
be controversial here, so I’m probably gonna
get stoned in a minute. (audience laughs) I never get through one of these things without breaking a little glass, and I’m surprised I ever get invited back. I think if you look at A.I.,
there’s kind of three areas. Number one, we have A.G.I., which is artificial general intelligence, and this would be the
Google Deep Mind Project, this would get to Kurzweil
and the singularity. This is the idea that
we’re gonna have computers that are equivalent to or
smarter than human beings. I don’t think there’s any
question that we can get a computer that can do any individual task better than a human. Perhaps drive a car,
play chess, play golf. But I think the idea that
we’re gonna computers that are gonna do all tasks
better than a human being any time in the future
or be more intelligent than a human being any time
in the foreseeable future in my opinion is highly unlikely, and so this Elon Musk rant. And by the way, Elon is
a very, very bright guy and someone for whom I have
enormous professional respect, and personal admiration, but I
do not think we need to worry about the malevolent computer refrigerator taking over the households, okay? It’s not gonna happen any time soon, and I think we underestimate
the capacity of the human mind and the human spirit and candidly,
I think the most powerful computer that we can conceive of today might have intellectual capacity
equivalent to a mealworm. Now, the second area of A.I. relates to the application of A.I. in
the field of social media. This is actually pretty troubling. Where we, through, where
social media companies have figured out how to use
computers to manipulate people, say, 2.2 billion at a time at
the level of an Olympic brain. So this has an Olympic brain,
it’s that part of the brain that we have that evolved
maybe 200 million years ago, and that’s what we have. That part of the brain, the
nervous system that we have in common with the
lizard, and it deals with the stuff of survival. What would that be? Food, water, sex, and it
releases a neurotransmitter called dopamine, which is the
pleasure center of the brain and this is how we learn to do things that result in survival of the species, like eat and drink and whatever else. Now, here we are definitely
manipulating A.I. in, I think, some pretty nefarious ways, and I think some of the
consequences of this on human beings and society
are pretty deleterious. That we have this whole concept of iGen for this post-Millennial
generation where we have massive depression, loneliness,
people don’t have friends, people don’t have dates,
people don’t have social lives, we have the weaponization
of these media by bad actors that call into question
whether or not we’re going to be able to conduct
a democratic society. So there’s some really
troubling things going on there. Then we have, so we get into, for those of you who know
about Siebel Scholars, we’re holding a Siebel Scholars
Conference on that topic at the University of Chicago next month that I’m really looking forward to. Now, the third general
area of A.I. relates to the application of A.I.
to commercial, industrial and government processes, and
there are classes of problems there that might seem relatively
mundane, but they were unsolvable as recently as 10 years ago, and they result in enormous
social and economic benefit. Let’s say A.I.-based
predictive maintenance. We can go into a power
grid infrastructure. Let’s talk about smart
grid analytics in general, where we can go to the grid
infrastructure, the grid being the most large and most
complex machine ever built and according to the National
Academy of Engineers, the most significant
scientific achievement of the 20th century. But now in the last 10 years,
we’ve been upgrading this grid infrastructure so
that all of the components of the grid infrastructure
are being upgraded so they’re remotely machine addressable. This is what we call IOD, right? So generation, transmission, distribution, transformers, substations,
reclosures, capacity mix, they’re all remotely machine addressable. So we can use A.I. to
aggregate all those data and these data sets are
massive and operating at a very high velocity,
with some of these data writing at, say, 90 hz cycles. We can run these data through
machine learning models at the rate that these data
arrive and dramatically increase the resilience, lower the cost,
reduce the amount of fuel, reduce the vulnerability
to cyber security, increase their reliability and
reduce the greenhouse gasses by a factor of two, so this
is something we go today in Spain, we do it in
Italy, we do it in France, we do it in New York. Digital oil fields. I know that some believe that in two years we’re all going to stop using
hydrocarbon products globally, but don’t bet on it, and the, we can in fact use A.I. in upstream, downstream, midstream
to increase productivity, increase safety, increase
reliability, lower the cost, human cost associated with injury, lower the human cost associated
with environmental impact by getting a environmental disaster. Other examples of A.I.
that are very common relate to financial services. So in the financial services
industry, fraud detection, anti-money laundering,
Volcker Act compliance, inter-day liquidity. It has to do with, it’s basically
just cast of optimization of the supply chains. Any inventory optimization problem. How much cash do I need where, when to meet the needs of the market? What the killer application in A.I., it would be A.I. based
predictive maintenance. Whether it’s a Caterpillar tractor, whether it’s a Boeing 737 or whether it’s a F-35
joint strike fighter. We can take all the
telemetry from these systems and run these telemetries
through machine learning models and predict system failure in advance, whether it’s propulsion,
auxiliary power unit, pressurization, flight
controls, and if we can predict the failures, say, 50 to
100 flight hours in advance, we can avoid unscheduled maintenance and increase the availability of aircraft and also dramatically decrease air accidents, and we
do this, for example, for the United States Air Force. The United States Air
Force has 4,600 aircraft and today we’re doing A.I.
predicative maintenance for E-3 Sentry, that’s the
AWACS, F-15, F-16, F-18, F-35 joint strike fighter, and we increase the availability of some of these aircraft from an order of 50% to an order of 80%, and if you’re in the business of being the Secretary of Defense, that’s
actually a pretty big deal, you know, when some of
these air crafts cost $1 million a copy, and they’re grounded. These are types of applications
that we do every day. What will the largest
application of A.I. be? Hard stop, precision medicine. We can and do, so it is within the state of the art to aggregate the
healthcare records of, say, the population of the United States into a unified federated image. Soon, this will encode
the genome sequence. So, what are things we can do there? We can predict across a population
of say 330 million people who is going to be diagnosed
with what disease in the next five years with very high levels
of precision, say diabetes. This gives us the opportunity
to, or heart disease, or breast cancer, or whatever it may be, and it gives us the opportunity
to intervene clinically now and avoid the diagnoses. How about genome specific
medical protocols? They’ll be much more
efficacious at much lower cost. Now, I’m not buying any of the Watson ads that A.I.’s gonna replace physicians. I don’t believe it, but unquestionably, A.I. is going to assist
physicians in diagnoses, whether it be this radiology
or whatever it may be. And so, adverse drug reaction. We’ve demonstrated that
we can take, we’ve taken the medical records for
40% of the United States, this in 125 million people,
aggregated those records into a unified federated image, and built a machine learning
model that will predict, with 85% precision and 80%
recall, that’s kind of A.I. talk for pretty accurate,
actually highly accurate, and predict, were we to prescribe opiates to this population, who’s
predisposed to addiction? So that will be, unquestionably,
the largest application of A.I. is precision medicine,
and it will lower the cost of medicine, we will live
longer and we will be healthier. – You know, your last comment,
Eric Topol has this book, I don’t know if you’ve seen it, “Deep Medicine,” which really,
he makes the same case. He’s a practicing, head
of the Cleveland Clinic, he’s now in Ohio. He makes the case that
deep learning will actually allow clinicians and
healers to spend time doing what they were supposed
to do, namely healing rather than all the stuff that’s needed. So anyhow, listening to you really, I think it reminded me of that. Okay, Jane’s telling me I
shouldn’t ask too many questions. We’ve got to get to the audience. But I want to ask you
one last one, Jane, okay? (audience laughs) Okay, there are a lot of other questions. Sorry, I couldn’t get to them all. One last one. Throughout the book. (laughs) I promise, I promise. (laughs) About the need to
revisualize the workforce, and your experience in this regard at C3. So share with us, because
there are major leadership action plans for us also in academia to prepare students for
this huge transformation. – Well, let’s talk about
the future of work. Will these technologies
have a disruptive impact on the workforce? Absolutely. Did the automobile have
a disruptive impact on the workforce? Absolutely. Did the jacquard loom, did
the Industrial Revolution? I mean, all of these technologies,
people were left behind who didn’t keep up. Now, again, this is going to have a significant impact on the workforce. I mean, when we have autonomous vehicles, what are taxi drivers gonna do? There’s kind of two ways that companies can prepare for this. They can get ready to fire
everybody and hire a new workforce or they can train
the workforce that we have. And I think the companies
who are really leaders in this field are leaning
on these technologies, many of them developed right
here, some of them developed at the University of Illinois,
MIT, Stanford, you name it, for continuous learning. At C3 today, we employ about 400 people, about 65% of our people
have advance degrees, and they have their
PhDs or Masters degrees in Data Science from UC
Berkeley, MIT, Carnegie Mellon, you name it, and they are obsolete two weeks after they show up to work. (audience laughs) Seriously. If you don’t have a book in your hand, if you’re not constantly
learning in the 21st century, you have a problem,
and so what we do at C3 and I think what many companies
will be doing is investing in their human capital to
learn the skills that we need. And, so for example, at C3, we’ve curated a curriculum from Coursera
in machine learning, deep learning, supervised
learning, Kubernetes, virtualization, elastic cloud computing, natural language processing,
besan processes, you name it, and we encourage people
to take these classes and get their certificates
and we pay them, I don’t know, 1,000 or $2,000 when they
get their certificate. And then we feature them in the newsletter and we feature their
names on the wall of fame in our headquarters
building, and now for people that wanna get Masters
degrees in Computer Science or Data Science, I know you
have a similar program here. There’s another program at another okay Computer Science program at
the University of Illinois, UC Berkeley, for those
people who want to get their advanced degree in Data
Science or Computer Science. They enroll and get admitted,
which is not that easy to get admitted, as you know,
and then we pay the tuition, we pay the fees, we provide
them a $25,000 bonus when they get their degree
and we get a 15% increase in cash compensation and
additional equity in the company. So, at C3 A.I., learning is
a core part of the culture. I think companies that really
thrive in the 21st century, learning will be a core
part of the culture to ensure that their
workforce has the skills to meet the needs of this
digitally transformed economy. – Fantastic. Alright, I promised. – I think we have time
for just a few questions. – Speak a little louder, I
don’t think the mic’s on. – [Man] Hi, welcome to Berkeley. – Thank you. – [Man] I know you touched
on early in this lecture how during the start of the
21st century there are a lot of companies that went out of
business that were big players in the 20th century. I’m wondering what you
think about how some of the big players, such as Amazon
and WalMart that are sort of expanding their reach and
their retail and market space, are possibly, what sort of
role do these major actors have on this company extinction
and the shrinkage of the number of companies in the market? – 8,000 retail outlets
have shuttered their doors in the last 12 months
in the United States. Big box stores, you name it. And I think Amazon looks
like a major disruptor in that place, in that space. I think if WalMart doesn’t
their act together, I think they’re kind of looking
down the barrel of a gun pointed at them from Amazon,
and if WalMart doesn’t get their act together, they’re
gonna be in a world of hurt. – [Woman] Thank you. You mentioned being trained
in History of Science and you’ve talked a lot
about the technology in relationship to human beings. I wonder what do you think
for today’s undergrad students or even employees in your
company, what do they need to know about history or about the relationship of technology and society? – I think people need to know
everything about history. Let me comment, I know
that it’s very vogue and I got in trouble with
Tsu-Jae in one of her meetings when I went off on a jag on this, so I know I won’t be invited back to that. (audience laughs) There’s this emphasis that
everybody has to study science, technology,
engineering and mathematics. I mean really, why? Why? What’s the matter with classics? How about literature? If you don’t understand history, how can you operate in this world? So I think all of these
things are important, and I think they’re equally as important as electrical engineering
and data science. Let me give you one example. We were working with Enel
doing predictive maintenance. Doing fraud detection for energy in Italy, and we had some of the
best data scientists on the planet Earth, David
Coulter from Carnegie Mellon, Andrew Olson from UC Berkeley. I mean, these guys are genuine big shots, breaking their picks building
machine learning models that would predict who’s stealing energy, how much are they stealing and how are they doing it
within any degree of precision. This is going on for six months. We have very high levels
of precision and recall in Sicily, like 85%, and every place else, it’s like 8%, which is, I can
assure you, not acceptable. And so I sat down with
these guys and I said, well, let me see the data. And they looked at, well,
all the true positives were basically coming from Sicily, and what they had done,
they were trying to build machine learning models for Italy, and they had built a
machine learning model that operated in I think
about 800 dimensional space. It was a supervised learning
model that worked for Sicily. And I said, have any of you
guys evert read history? Italy is not one country. Italy used to be hundreds of city states, and then it was 14 kingdoms. So I went on, you know, some
of you have read Machiavelli. You remember this stuff, right? And so I went on Wikipedia
that night and printed out a map of the 14 kingdoms of Italy. What year would that’ve been,
that we had 14 kings of Italy? – It might have been as
late as the 18th century. – Okay, so the 14 kingdoms of Italy, and I brought the map into the office, and the 14 kingdoms of Italy
happened to map exactly to the 14 regions of Enel, in Italy, in which they operate what they call DT, which are their operating regions. The problem is you don’t need
one machine learning model. You needed 14 machine learning models. Think about it: we have
the people in Sicily and the people in Venice. They have different genomes,
they have different diets, they have different forms of work and they steal energy differently, okay? (audience laughs) So, then that’s what they did. They built 14 different
machine learning models. It increased our precision
by an order of magnitude and completely solved the problem. So yes, I think history’s important. (audience laughs) (audience applauds) – Alright, thank you so much, Tom. I think we’ll have to
conclude the formal portion of our event today, but
before you leave Tom, we’d like to have our
students from Tau Beta Pi present you with a gift, a
token of our appreciation for spending time with us today. (audience laughs)
– Alright! I need that. Thank you. (audience applauds) Alright. – And to everybody in the audience, Tom has been very generous. He’s going to provide copies of his book, and he’s willing to sign them. Thank you so much, Tom. And we’re gonna provide
ice cream also outside, so please join us outside for a reception. And join me in thanking Tom
again for a wonderful talk. (audience applauds)

Leave a Reply

Your email address will not be published. Required fields are marked *