00:00:00
welcome welcome everyone to the first episode of the analytics at Wharton and AI at Wharton podcast series on
00:00:07
artificial intelligence my name is Eric bradow I'm a professor of marketing and
00:00:12
statistics here at the Wharton School I'm also Vice dean of analytics and I will be the host for this multi-part
00:00:17
series on artificial intelligence I can think of no better way to start that series with both two of my friends and
00:00:24
two colleagues who actually run our Center on artificial intelligence the title of this episode is artificial
00:00:30
intelligence is here as you will hear we'll do episodes on artificial intelligence in sports artificial
00:00:35
intelligence in real estate artificial intelligence in healthcare but I think it's best to start just with the basics
00:00:42
so I'm very happy to have joined with me today first my colleague cik hoser our
00:00:47
cardic is the John C Hower Professor um at the Wharton School he's also as I
00:00:51
mentioned the co-director of our Center on artificial intelligence at Wharton um
00:00:56
and normally I don't read someone's bio first of all it's only a few sentences
00:01:00
but I think this actually is important for our listeners to understand the breath and also the practicality of
00:01:05
cic's work um his research examins how AI impacts business and society and something you'll hear about is that is
00:01:12
what our Center does there's kind of two prongs second he was a founder of Yodo
00:01:16
where he applied AI to online advertising and more recently and currently to jump cut media a company
00:01:22
applying AI to democratize Hollywood and he also teaches our courses on enabling
00:01:27
Technologies and AI business in society so cardic welcome thanks for having me Eric I'm also happy to have my colleague
00:01:33
Stephano poni uh Stephano is the Sebastian sresky professor of marketing here at the Wharton School he's also
00:01:40
along with cardic the co-director of our Center on AI at Wharton and his research
00:01:45
examines how artificial intelligence and autom automation are changing consumption and society and similar to
00:01:51
cardic he also teaches our courses on artificial intelligence brand management and marketing strategies so Stephano
00:01:56
welcome thank you very much it's great to be with both of you so maybe CTIC I'll throw the first question out to you
00:02:02
um while artificial intelligence is now the big thing that every company is thinking about what do you see as well
00:02:10
first of all maybe even before what are challenges facing companies how would you even Define what artificial
00:02:16
intelligence is because it can mean lots of things it could mean everything from
00:02:21
taking text and images and stuff like that and kind of quantifying it or it could be generative AI which is kind of
00:02:28
a same side of the coin but a different part how do you even view what does it mean to say artificial intelligence yeah
00:02:35
artificial intelligence is a field of computer science which is focused on getting computers to do the kinds of
00:02:41
things that human that traditionally requires human intelligence and so what that is is the moving Target so when uh
00:02:50
computers couldn't play uh you know say a very simple game like um well chess is
00:02:55
not simple but uh you know maybe even simpler board games maybe that's the Target and then when you say uh
00:03:02
computers can play chess and when that's easy for computers we no longer think of
00:03:06
that as AI but really today when we think about what is AI it's again getting computers to do the kinds of
00:03:13
things that require human intelligence like understand language like navigate the physical world uh like being able to
00:03:20
learn uh from experiences from data so all of that really is included in in AI do you put any separation between what I
00:03:29
call I'll call it tra maybe I'm not even using the right words traditional AI
00:03:33
which again back in my old days we've had our AI around like how do you take
00:03:37
an image and turn it into something how do we take video how do we take text that's one form of AI versus what what's
00:03:44
got everybody excited today which is chat GPT which is a form of large language model do you put any
00:03:49
differentiation there or that's just a way for us to understand like one is kind of like creation of data and the
00:03:56
other one's kind of like using it in an application of forecasting language yeah
00:04:00
I I feel there is some distinction But ultimately they're closely related because what we think of as the more
00:04:06
traditional AI or predictive AI it's all about taking data and understanding the
00:04:12
landscape of the data and to be able to say in this region of the data let's say
00:04:15
you're predicting whether an um you know an image is about uh Bob or is it about
00:04:22
uh Lisa and so you kind of say you know in the image space this region if the shape of the colors are like this the
00:04:30
shape of the eyes are like this then it's Bob in that area it's Lisa and so
00:04:33
on so it's mostly understanding the space of data and being able to say with
00:04:38
emails is it fraudulent or not and saying which portion of the space does it have one value versus the other now
00:04:44
once you start getting really good at predicting that then you can start to use those predictions to create and
00:04:50
that's where it's the next step where it becomes generative AI where now you're
00:04:55
predicting what's the next word you may as well use it to start generating text
00:04:59
and start generating sentences essays and and novels and so on so Stephano let me ask you a question so if one went to
00:05:05
your website on the Wharton website and while by the way just for our listeners um Stephano has a a lot of deep training
00:05:12
in statistics but most people would say you're not a computer scientist you're
00:05:16
not a mathematician um what the hell do you have to do with artificial intelligence like what role does
00:05:22
Consumer psychology play in artificial intelligence today like isn't it just
00:05:26
for us math types so if you talk to companies and you ask them why did your analytics program fail
00:05:36
you almost never hear the answer because the model didn't work because the techniques didn't deliver it's never
00:05:43
about the technical stuff it's always about people it's about lack of vision
00:05:47
it's about a lack of alignment between decision makers and analyst it's about a
00:05:51
lack of clarity about why we do analytics so I think that a Behavioral Science perspective on analytics can
00:05:58
bring a lot of benef benefit to try to understand how do we connect decisions in companies to the data that we have
00:06:05
and so that takes both the technical skills and the human insights the psychology insights and so I think
00:06:12
bringing those together I find that has a a lot of value and a lot of uh you know potential insights that can a lot
00:06:19
of low hanging fruits in fact in companies I think as a follow-up question um you know we all read these
00:06:25
articles that say you know 70% of the jobs are going to go away and you know robots or automation or AI is going to
00:06:32
put me out of business should should employees be happy with what's going on
00:06:37
in AI or the answer is it depends who you are and what you're doing what are
00:06:40
your thoughts and then cardock I'd love to get your thoughts on that including
00:06:43
the work you're doing at Jump cut because we all know one of the biggest issues in the current writer strike was
00:06:49
actually what's going to happen with artificial intelligence so I'd love to
00:06:52
hear your thoughts from this psychology or the employee motivation perspective and then what are you seeing actually
00:06:56
out in the real world the academic answer to any question will be it depends it depends
00:07:02
but uh in my research what I've been looking at is the extent to which people
00:07:06
perceive automation as a threat and uh what we find is that often times when tasks that are being automated by AI for
00:07:15
example um are tasks that have some kind of meaning to the person that they are Central to the way that they see
00:07:21
themselves for example in their professional identity they can create a lot of threat so you have psychological
00:07:26
threats and then you have the objective threats of maybe jobs on the line and maybe you'll feel happy about knowing
00:07:32
that I try out the professor job on some of this scoring algorithms and we are fairly safe for now at least well and so
00:07:40
cardic let me ask you and let me just preface this with saying um You probably don't even know about this 15 years ago
00:07:46
I wrote a paper with a former colleague and a doctoral student about how to use it was I didn't call it AI back then but
00:07:53
how to basically in large scale compute features of advertisements and optim L design advertisements based on a massive
00:08:02
number of features and I remember the reaction I first thought I was going to get rich I went to every big media
00:08:08
agency and said you know you can fire all your creative people I know how to create these ads using mathematics and I
00:08:15
was looked at like I had four heads so can you bring us up to the year 2023 can you tell us what you're doing at Jump
00:08:21
cut and kind of what role AI machine learning plays in your company and just what you see going on in the creative
00:08:28
world yeah yeah yeah and I'll I'll connect that to also what you and Stefano just brought up about Ai and
00:08:34
jobs and and exposure to Ai and so on um I just came from a real estate conference and the panel before I spoke
00:08:43
was talking about hey this artificial intelligence it's not really intelligence it just uh replicates
00:08:49
whatever in some data we this true human intelligence is creative problem solving
00:08:53
and so on and I was sharing over there that there are multiple studies now that talk about what can AI do and cannot do
00:09:00
for example my colleague Daniel Rock has a study where he shows that just llms meaning large language models like Chad
00:09:08
GPT and before the advances of the last 6 months this is as of early 2023 they found that 50% of jobs have at least 10%
00:09:17
of their tasks exposed to llms 20% of jobs have more than 50% of their tasks exposed to llm and that's not all of AI
00:09:24
That's Just large language models and that's also 10 months ago and people all
00:09:29
so underestimate the nature of exponential change exponential and I've been working
00:09:34
with gpd2 gpt3 you know the earlier models of this and I can say every year the change is order of magnitude and so
00:09:43
you know it's coming and it's going to affect all kinds of jobs um now as of
00:09:47
today I can say that multiple research studies I don't mean two three four but
00:09:52
several dozen research studies that have looked at AI use in multiple settings including creative settings like writing
00:09:59
poems or problem solving and so on find that AI today already can match humans but human plus AI today beats both human
00:10:10
alone and AI alone so for me the big opportunity with AI is we are going to see productivity boost like we've never
00:10:18
seen before in the history of humanity and that kind of productivity boost allows us to Outsource the grunt work to
00:10:27
Ai and do the most creative things and derive Joy from our work now does that mean it's all going to be beautiful for
00:10:35
all of us no there are going to be some of us who if we don't res skill if we
00:10:41
don't focus on having skills that require creativity empathy teamwork leadership those kinds of skills then
00:10:51
lot of the other jobs are going away including knowledge work uh Consulting software development you know it's
00:10:58
coming into all of these so just to remind everyone this is Eric bradow professor of marketing and statistics
00:11:04
here at the Wharton School and also Vice dean of analytics we're here at the
00:11:07
analytics at Wharton AI at Wharton podcast series on artificial intelligence we're here with our first
00:11:12
episode of our at least 10p part series and this one is artificial intelligence is here and I'm talking to my colleagues
00:11:18
cardic hoser and Stephano pontoni so Stephano could um something cardic mentioned in his last thing was about
00:11:25
humans and AI matter of fact one of the things I heard you say from the beginning is it's not humans or AI it's
00:11:31
humans and AI how do you really see that interface going forward is it up to the
00:11:37
individual worker to decide what part of his her their task to kind of Outsource
00:11:41
is it up to management um how do you see it you know kind of how do you see people being even willing to skill
00:11:48
themselves up in artificial intelligence how do you see this I think this is a bigger qu biggest question that any
00:11:55
company should be asking not just about AI right now frankly I think the biggest
00:11:58
question of of all in business how do we use these tools how do we learn how to use them there's no template nobody
00:12:04
really knows how for example generative AI is going to impact different functions we're just learning about
00:12:10
these tools and these tools are still getting better so what we need to do is to have some uh deliberate
00:12:15
experimentation we need to build processes for learning such that we have individuals within the organizations
00:12:21
tasked with just understanding what this can do and it's going to be a um you
00:12:26
know impact on individual it's going to be impact on teams on workflows how do
00:12:30
we bring this in in a way that we just maybe not simply think of re-engineering a task to get the human out of the
00:12:36
picture but how do we re-engineer new ways of working such that we can get the most out of people the point shouldn't
00:12:43
be you know human replacement and obsolesence it should be human flourishing how do we take this amazing
00:12:49
technology to make our work more productive more meaningful more impactful and ultimately make Society
00:12:56
better so cardock let me take what what Stephano said in combining with something that you said earlier which
00:13:01
was about the exponential growth rate so my biggest fear if I were working at a company today and you please I'd love
00:13:08
your thoughts is that someone's using a version of chat GPT or some large language model or even predictive model
00:13:15
some Transformer model and they fit it today and they say see the model can't
00:13:21
do this and then two weeks later the model can do this and so companies in some sense create these absolute like
00:13:29
you just mentioned you were a real estate well chat GP or large language models AI can't you know sell homes they
00:13:35
can't build massive predictive models using satellite data yeah maybe they can
00:13:40
can't today but maybe they can tomorrow how do you in some sense try to help
00:13:45
both researchers and companies move away from absolutes in a time of exponential
00:13:50
growth of these methods yeah I think our brains fundamentally struggle with exponential
00:13:57
change um and probably there is some basis to this in you know studies people have done on Neuroscience or human
00:14:04
evolution and so on but we struggle with it and I see this all the time because I
00:14:09
have been part of that my work has been part of that exponential change from the
00:14:12
very beginning when I started my PhD it was about the internet and I can't tell
00:14:16
you the number of people who looked at internet at any given point of time and said nobody will buy clothing online
00:14:22
nobody will buy eyeglasses online nobody would do this nobody would do that and I'm like no no it's all happening just
00:14:28
wait to see what's coming and so I think it's hard for people to Fathom I think
00:14:33
leadership as well as Regulators need to realize what's coming understand what
00:14:38
exponential change is and start to work now for you brought up previously and I forgot to address it about like the
00:14:43
Hollywood writer strike now it is true that today chat GPD cannot write a great novel however when we work with writers
00:14:53
we are already seeing how they can increase the productivity for writers and how they can you know and in
00:15:01
Hollywood for example you know writers are notorious because they're you know
00:15:06
writing is driven by inspiration and you're expecting the draft today and what's the excuse oh I'm just stuck uh
00:15:13
at this point I you know um and when I get unstuck I'll write again and so you
00:15:18
can wait months and sometimes years for the writer to get unstuck and now you give them a brainstorming buddy and they
00:15:26
start getting unstuck and it increases productivity and yes they're right in
00:15:30
fearing that at some point they're going to keep interacting with the AI and keep
00:15:34
training the AI and someday the AI is going to say you know what I'm going to
00:15:37
try and write the script myself and when I say the AI is going to say that I mean
00:15:41
the AI is going to be good enough and some uh executive is going to say why deal with humans and and do that and so
00:15:47
I think we need to both recognize that change is that fast and start experimenting and start learning and
00:15:56
people need to start upping their game and res Skilling and get really good at using AI to do what they do and it's at
00:16:03
that reskilling is important stop viewing this as a threat because what's happening is you're standing somewhere
00:16:09
and there's a fast bullet train coming at you and you're saying that train is
00:16:14
going to stop on its own no it's going to run over you and the only thing you
00:16:18
can do and you have to do is get to the station board the train and be part of that train and help shape where it goes
00:16:25
all of us need to help shape where it goes mhm yeah one example I like to give is that um for I know 25 plus years I've
00:16:32
been doing statistical analysis in R and of course over the last 5 to seven years
00:16:36
Python's taken a much larger role and I always promised myself I was going to
00:16:39
Learn Python well I've learned python now I stick it directly into chat my rcode into chat GPT and I tell it to
00:16:45
convert it to Python and I'm actually a damn good python programmer now because
00:16:50
chat GPT has helped me take structured R code and turn it into python code that's
00:16:55
a great example and I'll give you two more examples like that the head of product at my company Jump Card media
00:17:01
had this idea for a script summarization tool so what happens in Hollywood is the
00:17:06
vast majority of scripts written are never read because every executive gets so many scripts and you have no time to
00:17:13
read anything and you end up prioritizing based on gut and relationships he Eric's my buddy I'll
00:17:20
read his script but not this guy Stefano who just sent me a script I don't know
00:17:23
him and that's how decision making Works in Hollywood so the head of product
00:17:27
who's not a he's actually a Wharton alumnist had this idea for a great script
00:17:32
summarization tool that would summarize things using the language and parland of
00:17:40
Hollywood and he had the idea to build the tool but he's not a coder our Engineers were too busy with other
00:17:47
efforts so he said while they're doing that let me try it on chat GPD and he
00:17:51
built the entire uh minimal viable product a demo version of it on his own using chat GPT
00:17:59
and it's actually on our website on jump cut media where our clients can try it
00:18:04
um and that's how it got built a guy with no development skills I actually demonstrated during this real estate
00:18:10
conference this idea that you've got you post a video on YouTube You've Got
00:18:14
30,000 comments on YouTube and you want to analyze those comments and figure out
00:18:18
what are people saying you want to summarize it I went to chat GPD and I said six steps First Step go to a
00:18:26
YouTube url I'll share download all the comments Second Step do sentiment analysis of that third step uh find the
00:18:34
comments which are positive and send it to open Ai and give me the summary of all the positive comments fourth step
00:18:41
negative comment send it open get the summary fifth step tell the marketing manager what you should do and give me
00:18:47
the code for all this it gave me the code in the conference with all these people I put it in uh Google collab ran
00:18:54
it and now we've got the summary and this is me writing not a single line of
00:18:59
code uh with ch gbd it's not the most complex code but this is something that
00:19:04
previously would have taken me days and I would have had to invol aray and so on
00:19:07
and I can get that done and also Imagine in real estate doing that about a property or a developer or and you say
00:19:13
it doesn't affect real estate of course it does absolutely it could it does and
00:19:17
I also showed them I uploaded four photographs of my home nothing else four photographs and I said I'm planning to
00:19:23
list this home for sale give me a a real estate listing to post on Zillow that would make people read it and get
00:19:31
excited to come and tour this house and it gave a a great beautiful description there's no way I could have written that
00:19:38
I challenged them how many of you could have written this and everyone at the end was like wow I was blown away and
00:19:43
that is something that is doable today I'm not even talking about this is coming soon so Stefano let me ask you um
00:19:49
I'm G to ask you and then I'll ask cardic as well what's at the Leading
00:19:54
Edge of the research you're doing right now so I want to ask each of you about
00:19:57
your own search and then I'll spend the last few minutes that we have talking
00:20:01
about AI at Wharton and what you guys are doing and hoping to accomplish so let's start with your own personal
00:20:06
research like what are you doing right now or another way I like to frame it is if we're sitting here 5 years from now
00:20:11
and you have a bunch of published papers and you've given a lot of big Podium
00:20:14
talks which I know you do what are you talking about that you had worked on um working on a lots of projects all in the
00:20:21
area of AI and so many exciting questions because we never had a machine like this a machine that can do the
00:20:26
stuff that we think is crucial to defining what a human is this is actually an interesting thing to
00:20:32
consider when you went back in time maybe a few years and you asked what makes human special people were thinking
00:20:37
you know maybe compared to other animals we can think and now we ask what makes human special and people see instead oh
00:20:45
we have emotions or we feel and so basically now what makes us special is what makes us the same as other animals
00:20:50
to some extent so you see how the world is really deeply changing and I'm interested in for example the impact of
00:20:57
AI for the pursuit of relational goals or social goals or emotional um heavy type of tasks where previously we never
00:21:06
had an option of engaging with a machine but now we do and what does that mean um
00:21:10
what are potentially the benefits that this technology can bring but also what might be the dangers for example for
00:21:16
Consumer safety as people might interact with these tools while experiencing mental health issues or other problems
00:21:22
that's to me that's a very exciting and important area and I just want to make a
00:21:26
point that this technology doesn't have to be any better than it is today for it to
00:21:31
change many many things I mean Kik was saying rightly this is still increasing exponentially and companies are just
00:21:37
starting experimenting with it but the tools are there this is not a technology around the corner is in front of ush so
00:21:45
CK what what are kind of the big open issues that you're thinking about and working on today yeah Eric there are two
00:21:51
aspects to my work one is slightly more technical um and the other is focused more on humans and societal uh
00:21:58
interaction with u with AI so on the former side uh I'm spending a lot of time thinking about biases in machine
00:22:03
learning models uh in particular a few studies related to biases in text to image models for example you go in and
00:22:10
you write a prompt uh generate an image of a child studying astronomy if all 100
00:22:16
images are of a boy studying astronomy then you know there's an issue and and
00:22:20
and these models do have these biases just because the training data sets have that so but if I get an individual image
00:22:27
how do I know is it's okay or not and so we're doing some work on detecting bias
00:22:33
debiasing on automated prompt engineer engineering as well so you've you state
00:22:38
what you want and we'll figure out how to structure The Prompt for a machine
00:22:41
learning model to get the kind of output you want so that's a bit on the technical side on the human and AI side
00:22:49
uh most of my interest is around two themes one is human AI collaboration so if you look at any workflow in any
00:22:55
organization where AI now can touch uh that workflow we do not understand today what is ideally done by humans and what
00:23:03
is done by uh AI you know in terms of organization design and process design we understand historically like for
00:23:11
example how to structure teams uh how to build Team Dynamics but if the team is AI and humans you know how do we
00:23:19
structure that what should be done by whom so have some work going on there and the other one is around trust you
00:23:24
know AI is a huge trust problem today we were just talking about the right strike there's an actor strike and many
00:23:30
more issues coming up so what does it take to drive uh human trust and engagement with AI is another theme that
00:23:37
I'm looking at so maybe in the last few minutes or so Stephano could you tell us
00:23:41
a little bit and our listeners here on SiriusXM and on our podcast about AI at Wharton and what you're hoping to study
00:23:49
and accomplish through a center on artificial intelligence here at Wharton and then we'll get card's thoughts as
00:23:53
well yeah and thank you for organizing this podcast series for having us I think it's a great opportunity to get
00:23:58
the word out the um initiative AI award is just starting out we are you know a bunch of academics working on AI
00:24:06
tackling AI from different angles for the purpose of uh understanding what it can do for companies how we can improve
00:24:12
decision making in companies but also what are the implication for all of us as workers as consumers and Society
00:24:18
broadly so we're going to try to you know initiatives around education around
00:24:24
research around dissemination of research findings and try to create a community of people who um you know are
00:24:31
interested in these topics they're asking similar questions maybe in very different way and then we can learn from
00:24:37
one another and cardic what are your thoughts about you know why you know you've been involved with lots of
00:24:42
centers over the years what makes AI at Wharton special and why are you so excited to be in one of the leadership
00:24:47
positions of it yeah I think first of all to me AI is a one maybe not even a once a
00:24:56
generation but one several generation kind of Technologies and it's going to
00:25:01
open up so many questions that will not be answered unless we create initiatives like ours
00:25:08
for example today computer scientists are focused on creating new and better models but they're focused on these on
00:25:17
assessing these models somewhat narrowly in terms of accuracy of the model and so
00:25:21
on and not necessarily human impact societal impact you know some of these other questions at the same time
00:25:29
industry is affected by a lot of this but they're trying to put the fire out
00:25:34
and they're focused on what do they need to get done this week next week they're
00:25:37
very interested in the questions of where will this take us three four years later but they have to focus quarter by
00:25:43
quarter and I think we are uniquely positioned here at Wharton in terms of having both the technical chops to
00:25:51
understand those computer science models and what they're doing as well as you
00:25:56
know people like Stefano and others who understand the psych psychological and the social science Frameworks who can
00:26:03
bring in that perspective and really took a take a 5 10 15 25 year kind of timeline on this and figure out what
00:26:10
does this mean for how organizations need to be redesigned how does what does this mean in terms of how people need to
00:26:17
be reskilled how do our own college students uh need to be reskilled what does this mean for regulation because
00:26:24
man Regulators are going to struggle with this and while the techn teolog is moving exponentially Regulators are
00:26:29
moving linearly and so they will need that thought leadership as well so I think we fill that Gap uniquely in terms
00:26:35
of those kinds of problems big open issues that are going to hit us in 5 10 years but we are currently too busy
00:26:43
putting out the fires to worry about the the big Avalanche coming our way well I
00:26:47
think anybody that has listened to this episode will agree artificial intelligence is here which is what the
00:26:53
title of this episode was um again I'm Eric bradow professor of marketing and
00:26:57
statistics here at the Wharton School in Vice of analytics I'd like to thank my
00:27:01
colleagues Stephano ponton and CTIC ker thank you for joining us on this episode
00:27:05
thank you edic thank you