00:00:00
welcome to the next episode of the analytics at Wharton and AI at Wharton series on artificial intelligence um
00:00:08
this series is or this episode is probably one of the topics that we should be spending a lot more time
00:00:13
thinking about which is AI and Automotive in a lot of ways it's really Where it All Began I'm honored to have
00:00:20
my friend and colleague John Paul McDuffy speaking to us today uh John Paul is a professor in the management
00:00:25
department he's Al the also the director of the program on vehicle and mo IL
00:00:30
Innovation I'm going to want to ask him what that is as part of the Mac Institute of innovation management here
00:00:34
at the Wharton School uh so John Paul Welcome to our podcast today thank you glad to be here yeah so why don't before
00:00:40
we get into Ai and Automotive which is the big topic of today why don't you tell us I I mean I've always thought of
00:00:46
you as our innovation in automobiles guy so what is the program on vehicle and Mobility Innovation and what kind of
00:00:53
things do you guys doing probably the best shorthand for our research agenda is is the so-called
00:01:01
case technologies that are transforming Mobility c c for connected a for autonomous s for
00:01:09
shared Mobility business models and E for electric and understanding each of those Technologies and their
00:01:18
implications but also how they combine because they certainly combine in some cases and in other cases not um we for
00:01:27
example think that most autonomous vehicles going forward are likely to be electric but obviously not all electric
00:01:32
vehicles are fully autonomous and uh it's a fascinating competitive space we
00:01:38
have not only the Legacy incumbents we have newcomers like Tesla we have a lot of other newcomers you may not have
00:01:44
heard of and then we have big Tech hovering on the edges you know apple and and Google
00:01:49
slwo um uh foxcon you know Apple's manufacturer wants to get into the autonomous vehicle business and uh you
00:02:00
know run an entirely different you know iPhone type model for how this industry evolves so there's just uh a lot of
00:02:09
fascinating things going on another thing to say about um the history of this program on vehicle Mobility
00:02:14
Innovation I got my start on all this at MIT as a doctoral student with something
00:02:19
called the international motor vehicle program at the time it was trying to understand the transformation from
00:02:24
traditional mass production which had dominated most of the 20th century starting with Henry Ford to um Toyota
00:02:30
production system which uh our program gave the name lean production to so why was lean production supplanting mass
00:02:38
production why did it have competitive advantages and everything from manufacturing and product development to
00:02:43
Supply Chain management and the like and when I came to Wharton um I continued an
00:02:48
affiliation with MIT for a long time uh even was co-directing that program from down here uh at a certain point that
00:02:56
program was about to close down at MIT and I actually asked if I could move it to
00:03:02
Wharton and they said sure because uh at this point the program was really a network of Automotive researchers all
00:03:08
over the world that we kept kind of loosely coordinated sometimes we would simply get together and share knowledge
00:03:13
sometimes we would do joint Global research projects together and so to move kind of the network Hub to Wharton
00:03:21
from MIT was not such a big deal and uh program on vehicle Mobility Innovation pvmi is the opposite set of uh order the
00:03:29
initials of imvp that's a little in joke for those of us in the program I like
00:03:34
that one so let me ask you um when I think when I say historically I don't mean back 20 years when I think over the
00:03:41
last s to 10 years I really do as I opened up our podcast about I think of AI and Automotive as being like that was
00:03:49
the flagship like that was going to demonstrate the ability of AI to you know self-driving cars I guess as you
00:03:57
may call it you put in your notes level four where do things stand today like is
00:04:01
that dream are we going to be having driverless cars all over the place soon is level four autonomy even the goal
00:04:08
anymore yeah these are great questions and I think anybody who's followed this
00:04:12
technology and has had as in in interest in it knows about the hype in the sort of 2016 17 maybe even a little earlier
00:04:21
phase and how it's been kind of disappointing since then uh and yet these days you hear about uh Cruz and
00:04:28
weo and Zo operating you know on call Robo taxis in San Francisco on a limited basis um you may also hear about some of
00:04:37
the controversy with uh fire truck collisions and the regulation that's on the books in California to limit this
00:04:44
one of the things I always think about is like you know I used to work for a large I could say dupond large
00:04:49
International Chemical Company and like an error rate of one in a thousand one and 100,000 one and a million might be
00:04:56
fine what could possibly be an acceptable error rate for a product and service like this like do you have to be
00:05:03
like 10 standard deviations out on the safety scale like how do you even launch a product where one in 100,000 error
00:05:11
rate which is great for most products and service would be totally unacceptable here I mean I think the the
00:05:17
right answer the candid answer is we really don't know and it of course depends partly on public perception and
00:05:23
what people feel is safe it depends on what how Regulators think of this issue and of course on the progress of the
00:05:30
technology you know the proponents uh who say this is already a much safer technology would say you know 40,000
00:05:39
people a year die in car accidents in the US those are all cars driven by humans right and that after about 50
00:05:46
years of that rate going down in the last four or five years it's gone up despite all the new safety technology um
00:05:55
probably distracted driving in phones is is one of the big reasons there seem to
00:05:59
be two things going on at the same time when a user consumer has a first experience of a driverless car they're
00:06:07
like nervous for a little while then they ask a lot of questions and then pretty soon they get bored and they
00:06:12
start looking at their phone in other words people adapt to the experience very quickly and once they decide it's
00:06:17
basically looks like it's driving normally they don't worry about it but
00:06:22
whenever there's a big visible accident and the biggest one was the 2018 death
00:06:28
of a pedestrian Wheeling a bike by an Uber car which actually had a human operator there to keep an eye on the
00:06:36
autonomous vehicle she was actually looking at a phone at the time um that had a dramatic impact on how the public
00:06:43
felt and then suddenly you had a spike and people who said I would never even get in and try one of those and it put a
00:06:50
chill on many other aspects so I think it's still true of a lot of Technologies
00:06:55
we accept less error from a automated systems than we do from humans because we feel like we kind of understand the
00:07:03
sources of human error and particularly with AI you know behind the choices that
00:07:08
are being made with automated driving we don't feel like we understand those
00:07:12
choices well let me ask something that our listeners may be interested so we all know today generative AI is the next
00:07:17
big thing or at least people are talking about it chat GPT Etc if you if I think
00:07:21
about it as a statistician which I am um I think it's just a extremely high-dimensional prediction problem
00:07:28
isn't Ai and Automotive just a very high-dimensional prediction problem and so is there any reason why the I'll call
00:07:36
it General advances in AI today which tend to be focused right now on prompt engineering and chat GPT won't those
00:07:45
advances eventually help Ai and Automotive as well or do you see them as different problems you know I always say
00:07:53
what we do as academics is while the jargon changes and ones like what does John Paul McDuffy type into a chat box
00:08:00
and what comes back and the other is what does weo build into some automated AI driven vehicle to me they're both
00:08:06
high-dimensional prediction problems or do you see it differently no I I think the underlying technology affects both
00:08:13
and the progress of the underlying technology affects both um they obviously you know where where words is
00:08:19
the primary uh you know coin of the realm as it is with gen versus driving decisions there there's there's some
00:08:26
differences but you know I think part of what happened with autonomous driving is
00:08:32
the progress was remarkably rapid at first it started with Google competing successfully in a DARPA you know a
00:08:40
government defense department I teach that DARPA challenge in my MBA class it's a great example and it was everyone
00:08:46
was astounded at how well the Google software combined with about half a million dollars of hardware on the
00:08:51
vehicle did and so that fueled the the VC and all the other funding and and everyone getting into it let's imagine
00:08:59
progress that was rapid up to 90% of handling driving situations 95% 97% wow this thing is going to be
00:09:07
everywhere in no time at all you hit that last couple of percent you know I can't say exactly what it is the
00:09:13
so-called Corner cases the very rare combination of events that are really hard to you know either teach with
00:09:24
traditional programming here's what may happen and here's what you have to do or
00:09:27
even teach with the more inductive you know way of of machine learning which is you have a lot of data that you train it
00:09:33
on right these very rare events by definition don't happen very often and how do you how do you train them you can
00:09:39
write simulations and that's what these companies are doing okay let's imagine a
00:09:43
weird situation or let's find a freak accident that happened in the real world
00:09:48
and let's create a create a simulation for that their confidence that they've
00:09:53
figured out how to handle all those simulated situations as part of where they say yeah we're going to be able to
00:09:57
conquer it all but for the public and for regulators and all of us who think about you know climbing into these
00:10:03
vehicles uh every time one of those weird situations is not handled well it adds doubt to that side of the ledger so
00:10:13
um and I don't know that anybody ever feels that AI is going to solve everything 100% we're on warning right
00:10:20
that there's hallucination and there's all these problems with Gen that means
00:10:23
we have to really be alert this is Eric bradow professor of marketing statistics
00:10:28
and data science here at the Wharton School and also Vice dean of analytics and we're here as part of our analytics
00:10:33
at Wharton and AI series uh we're talking to professor John Paul McDuffy we're talking about Ai and automation
00:10:39
today so let me ask you do you think the future is that maybe Ai and automation will be more widespread but it might not
00:10:46
be what your just what you call level four like probably the next step for there to get more mass adoption is I I
00:10:53
guess it would be level three which would be yes self-driving but there might also be a steering steing wheel or
00:10:59
there might be the opportunity for human intervention or what do you see is going
00:11:03
to take it from let's call it what it is now which is really a niche market to
00:11:07
something that might be more widespread yeah a brief background on those levels it's um something the Society of
00:11:14
Automotive Engineers came up with to describe different levels of autonomy uh level there's actually a fifth level
00:11:22
which is can go anywhere at any time in any circumstances without being able to connect to the internet um that's even
00:11:29
further out but level four is basically in most operating conditions can operate
00:11:34
autonomously you're right level three involves some handing back of control between a human driver and an automated
00:11:40
system level two is stuff that we already see in a lot of modern vehicles Lane control selfing L control even
00:11:47
automatic braking I mean some of it's becoming becoming pretty Advanced so one
00:11:53
kind of strategic I don't know if it's a full divide but it's something you can
00:11:56
see the level for Stuff got all the headlines and that's what weo is investing on and that's what Cruz some
00:12:03
of the other prominent startups that's what Tesla with Alon musk has been promising for years with his you know
00:12:08
optimistically or misleadingly named autopilot system um the Legacy automakers and all the suppliers who
00:12:15
feed in the technologies have been slowly adding the level two advanced stuff and even experimenting a little
00:12:22
with level three and they can say we're making a safer car they can usually charge extra money for it although
00:12:30
increasingly a little bit following toa's lead there's a tendency to bundle
00:12:34
all the safety stuff together and just say this is the right thing to do buy it you know you get it all at once basic
00:12:40
prospect theory you you bundle prices together and so it's kind of a question of whether
00:12:48
the slow moving up from the lower levels ends up affecting more people's Driving Experience sooner than the
00:12:57
promise of level four which may stay a niche until it satisfies a lot of the questions we have about it so who's
00:13:03
going to be you know I've thought about this matter of fact I teach an automotive cases I mentioned in my in
00:13:07
the NBA Corp because I I think it's such a fascinating industry who's going to be
00:13:11
the winner here and let me even put on my so besides being Vice of analytics I am the chair of the market I'll put on
00:13:16
my marketing department hat here who would I trust more to get into an autonomous AI driven vehicle would I
00:13:24
trust Ford or would I trust Google and to me I'm thinking I want a data company I want a company
00:13:31
that's really good in Ai and Predictive Analytics I mean like another way to
00:13:37
frame it is do you ever see a day where the Legacy automakers actually turn out to be the
00:13:43
big winners in this or is it likely to be a tech company that just happens to also do work in
00:13:50
automotive well I mean that's a great and big question and so we may want to
00:13:54
spend a little time on it yeah well that's what we're doing here on the AI
00:13:57
series Ai and we're talking about big questions the um weo which is the Google
00:14:03
subsidiary doing this uh they've said pretty clearly we are not going to build
00:14:08
a vehicle we're not going to have a vehicle we're making a software driver
00:14:12
which we're going to sell to license to the people who make the vehicles so you
00:14:18
could have a wayo driver in a Ford vehicle maybe that would be a sweet spot for you if you like uh or or any other
00:14:26
Legacy automaker um you have other companies who are taking a different approach Tesla obviously being one but
00:14:33
zuk's is a company that is now owned by Amazon which is building the vehicle
00:14:38
hardware and the software and the business model for they have a robo taxi and also a automated Trucking like like
00:14:46
small Trucking almost delivery makes sense if they're own owned by Amazon um
00:14:52
so you have all these different combinations I've been increasingly feeling like even though the Automotive
00:14:59
companies and the tech companies don't really like each other and don't really
00:15:02
want to work together that they may have to there's something about Mastery of
00:15:06
the physical realities of a vehicle that the digital Giants really don't have and
00:15:10
obviously the auto companies are not good at digital stuff anybody who has tried the the company provided you know
00:15:17
interface in their car knows knows that for sure and I I think you know we instinctively think that the the the the
00:15:24
data Giants will be better at something that involves so much data that involves
00:15:28
Co connectivity and the like um and they've handled the control of physical systems braking and steering and the
00:15:36
basic stuff really quite well but um I think uh another aspect of this is the computer industry It generally has been
00:15:44
a quite a modular industry you have kind of clear interfaces you have Innovation
00:15:51
that's possible when people simply know the the specs the interface specs the
00:15:55
apis of the people they're working with the automotive industry has remained a
00:16:00
very integrated industry there's a lot of interdependencies in this complex multi-technology vehicle that can't be
00:16:08
predefined away it requires a lot of interaction to work it through and and this maybe is a clue to what you know
00:16:17
autonomous vehicles but also autonomous and electric and all the other stuff together it will require combining that
00:16:23
knowledge of the physical with the knowledge of the digital and will force some collaboration that maybe the
00:16:29
parties wouldn't choose otherwise so what's I don't use the word preventing
00:16:34
but let's use that what's preventing things from getting to even level five
00:16:39
autonomy is it um it would just cost too much to build that's one possibility
00:16:44
like you could build it but at that price just there would be such a small Market another possibility is you know
00:16:50
I'll say it we can't process like it's our Computing like we can't process like
00:16:55
something in a vehicle can't process that much information at that quick a speed it could be um we don't know
00:17:03
enough math it could be our algorithms aren't good enough yet what is the big
00:17:07
stumbling block given you said 95 plus percent was done in the first couple years what's preventing this last 3% cuz
00:17:17
I think I'm ready like I'm ready for level five no I'm saying I'd get into
00:17:20
level five sure I would yeah um there are new Chips coming along you know Tesla's designing it own chip
00:17:29
Nvidia it turns out the video processing chips are very good for the so-called Sensor Fusion Sensor Fusion is what they
00:17:36
call where you take the camera data the regular radar data and the lar the laser
00:17:42
laser radar data and you combine them to get that 360 picture of what's really
00:17:46
going on you have to have both accurate information about distance things are away from you and what they are right
00:17:52
you need to know is this a car is this a bike is this a person is this a a traffic you know piece of construction
00:17:58
equipment whatever see when you say that I have to admit if this wasn't professor
00:18:02
John Paul McDuffy of the Wharton Management Department this was just a general AI lecture I'd say so this could
00:18:08
be a store using facial recognition or this could be I mean just if you even think about the language you just use
00:18:15
this is the problem of AI engines today and it just happens the application area
00:18:20
in me in my view in automotive is really interesting and cool and possibly lifechanging yeah so you know and and
00:18:27
you were asking the various constraints on it yeah they're making progress on
00:18:32
the technical constraints um for sure there's the regulatory issue of how much
00:18:39
testing in real life to allow so as I said there's a lot of simulated testing
00:18:43
going on uh most of the testing is on a rather small scale so far it's rolling
00:18:50
out kind of state-by-state City by city for the robo taxis um there's also people working on automated Trucking and
00:18:58
uh trucks are much bigger which potentially a I mean people don't know this I teach this also in this same
00:19:02
lecture maybe you'll correct me wrong I've made the claim that the trucking
00:19:06
industry is the biggest industry in the US today um that's interesting you're
00:19:12
obviously combining every everything that moves Goods at any levela I don't know if if uh but what I've also
00:19:19
commented on is let's imagine level five automated truck driving happens I also
00:19:24
think about all the ancillary Industries like you know truck drivers gener a lot
00:19:28
of income for hotels and restaurants and everything else so we have a long conversation about let's call it related
00:19:34
Industries like if Ai and Automotive takes off there's so many other industries that would be impacted some
00:19:41
positively some not yeah you know there's a company named Aurora that's working um it it it took over the
00:19:48
autonomous vehicle stuff that Uber was doing it's working closer with Toyota
00:19:52
they're focusing on trucks right now and they're focusing on Long Haul uh
00:19:57
distances I think most people imagine that that is the part of trucking that's
00:20:03
the best application the local stuff which is the delivery where you're in cities and you're having to stop and put
00:20:10
things on people's doorsteps or put it you know in a in a some kind of box that's actually a lot more complex to do
00:20:17
right um and those those are also the jobs that are more local I mean they're
00:20:22
huge shortages of truck drivers the Long Haul truck driving life is not a very desirable occupation in terms of being
00:20:28
away from home and and health and the like and so but it it it reminds me of another distinction I wanted to make you
00:20:35
know we don't know if the autonomous vehicle is going to be a individual ownership model or it's only going to be
00:20:41
a Fleet model yeah that's what that get to affordability and that gets to you
00:20:45
know how do you manage the upkeep and maintenance uh you know Uber famously under its founding CEO Travis clonic
00:20:54
said you know this is existential for us to have to be able to move to a autonomous vehicles because we can't
00:21:00
afford to do the human driver model and achieve all the goals we want what's
00:21:06
your what's your forecast am I am I going to be owning an autonomous driving
00:21:10
vehicle or is there just going to be a fleet of vehicles out there that drive up to my home anytime I want it yeah I I
00:21:16
think the the full autonomy model probably works economically best as a Fleet model um you know remember though we
00:21:28
talked about the automation creeping up from level two through level three so level three is probably still a
00:21:34
personally owned vehicle right so um so then is there a niche for somebody who really wants their own vehicle that they
00:21:41
can then not drive or or or or not or will or will it all be or will it all be supplied now
00:21:48
there's also the economics of density and so if you live in a place where you
00:21:53
can guarantee relatively rapid service from a robot taxi you would need to but if if you're way out the country you're
00:21:59
probably not going to have that ready service available so then maybe you are a candidate to be the one who owns it
00:22:05
maybe you have to be really wealthy to afford it because the scale is not in the personal ownership it's in the it's
00:22:10
in the fleet um so you know the the the the there's some other issues about Fleet the companies that are doing the
00:22:21
software for these vehicles they don't know from running fleets and and repairing vehicles and dealing with Duty
00:22:28
Cycles and cleaning up after the last user and stuff like this so the minute they contract that out that's a chunk of
00:22:34
their profits but that's uh that's another story so in the last like two or
00:22:38
three minutes we have let me ask you um first it's a two-part question but it's
00:22:42
really the same question what are the open research questions from you as an academic in
00:22:48
this area and secondly um maybe we'll make a date 10 years from now you and I
00:22:53
are sitting here what are we going to talk about has happened over the last 10 years in the AI and a Automotive space
00:22:59
one of the big research questions I'm looking at is really about the the organization and structure of
00:23:05
the industry and of competition in this in this space so it's clearly expanding
00:23:09
Beyond Automotive to be a Mobility kind of space it's clearly Beyond just firm
00:23:15
competition to be ecosystems and what's going on with ecosystem competition uh there is a vision of
00:23:23
where this Mobility stuff goes which is around uh very modul systems uh open source software uh we haven't talked a
00:23:32
lot about it because it hasn't shown up so much yet but um for example I mentioned foxcon before uh they've
00:23:38
organized a Consortium of firms suppliers and the like they want to have uh open source
00:23:45
software for autonomous vehicles they want a contract manufacturing model for making the vehicles they want a
00:23:51
completely commoditize vehicle design where you start with a basic skateboard and then you put different kinds of and
00:23:58
it's it's right out of the digital Playbook of what we've seen happen in it
00:24:03
uh there are people who fervently believe in that and part of what they say is it's too damn expensive for every
00:24:09
single company to come up with its own vehicle its own software for you to have a competition even between wh Mo's
00:24:16
software and Apple's software for example is just a lot of why not have one open source that can also maybe be
00:24:24
vetted and and and that we can have more oversight than we would from a private big tech company um my basic skepticism
00:24:33
around this big research question is this fundamental fact that the architecture of both the product and the
00:24:38
industry has been up until now much more integrated because of all these kind of
00:24:43
interdependencies some of which are really based in physics and and physical realities I don't expect that to change
00:24:51
so much that this modular Vision really has the potential to happen that some of
00:24:56
its it or big Tech proponents would would like so that's kind of one big research question I'm looking at I guess
00:25:02
in 10 years we might be looking to see what the outcome is because there are some people who think that even if what
00:25:08
I said is true the brains are going to be somebody's software right somebody's
00:25:14
software is going to be so good that it wins yep and it's either a monopoly or
00:25:19
maybe it's an oligopoly you know maybe we end up with Android versus iOS you
00:25:23
know maybe there's just two autonomous driving you know and and electric and
00:25:29
everything uh competitors in the world and everyone else just has to bow down and you know uh oh yes you know I used
00:25:37
to be BMW but now I'll make your car you know oh oh oh oh big Tech Overlord um
00:25:43
who knows that uh may be possible remember that this is a super competitive super Global and super low
00:25:50
margin industry and these big tech companies are not used to that and I sort of question when I think about
00:25:56
Apple having an Apple car are not whether they could do it and do a very good job at it but whether they really
00:26:02
want to be in that business right and even if they pull back to say we just want to do the software um even that is
00:26:08
a big Challenge and we'll pull them away from a lot of the other things that
00:26:12
they're doing way Mo's committed to it I think weo will will stay in you've got
00:26:15
to be well financed so we may have a couple winners but it wouldn't surprise
00:26:20
me given what I've said before that you've got a couple of um essentially
00:26:26
alliances between y Auto and big Tech some of those alliances Thrive and win some of them fail and lose so we have
00:26:33
kind of a ShakeOut that leaves a few Legacy automakers and big Tech contenders on the floor and a few
00:26:39
winners that have figured out how to combine their uh complimentary capabilities let's say well John Paul
00:26:45
I'd like to thank you for joining me today on the analytics at Wharton AI at
00:26:49
Wharton sirusxm podcast series obviously we've been talking about AI in autom in
00:26:54
Automation and automotive and what's amazing to me was was I'm sure there
00:26:58
were times in your career where you've thought I'm in this old Legacy industry
00:27:02
but now's not one of them it's got to be a great let's be honest it's got to be a
00:27:05
great time to be you right now come on it's a great time to be me I'll say it
00:27:10
I'll agree well thank you for joining us and this uh again this has been Eric
00:27:14
bradow professor of marketing and statistics and data science byen here of analytics at the Wharton School uh
00:27:19
please stay with us and join us for our next episode of the analytics at Warden AI at Warden AI
00:27:26
series