00:00:00
we grow up with the movies and you know
00:00:04
and books and that highlight just the
00:00:07
dangers of Technology you know from The
00:00:09
Matrix Terminator and 2001 Space Odyssey
00:00:13
or blade done or whatever it is and
00:00:16
those kind of stories do inform
00:00:20
um how we think about technology moving
00:00:22
forward and especially today that we
00:00:24
almost seem to be living in a Sci-Fi
00:00:25
movie in a way where every day we read a
00:00:28
newspaper and we just thinking wow I
00:00:30
could never thought this would happen
00:00:31
welcome to the ripple effect the podcast
00:00:34
that takes you on a journey through the
00:00:36
minds of work and faculty I'm your host
00:00:38
Dan Looney and in each episode we'll be
00:00:41
diving deep into the inspiration behind
00:00:43
the groundbreaking research that Wharton
00:00:45
professors have conducted and exploring
00:00:48
how their findings resonate with the
00:00:49
world today we'll be covering a diverse
00:00:52
range of topics bringing you the latest
00:00:54
insights and knowledge that you can
00:00:56
apply to your life into work so get
00:00:59
ready to dive into new ideas with the
00:01:01
ripple effect
00:01:03
Stefano you've done a lot of
00:01:05
conversations and and papers on this
00:01:08
topic what was it that kind of Drew your
00:01:11
attention uh to this area of consumer
00:01:14
Behavior
00:01:16
I've been doing work on consumer
00:01:18
decision making for over two decades now
00:01:21
and I've been fascinated by the topic of
00:01:23
consumer Behavior because it's such a
00:01:26
big part of what we do every day you
00:01:28
know small decisions like whatever
00:01:30
buying a coffee or toothpaste but also
00:01:33
big decisions like you know buying a
00:01:36
house or or a car and I think you have a
00:01:39
big role to you know to to play in our
00:01:43
well-being in the way we live our lives
00:01:45
reflects a lot of things about who we
00:01:47
are and what we want to be and so I
00:01:49
think understanding consumer behavior is
00:01:51
a very interesting fascinating
00:01:52
fascinating lens into human psychology
00:01:55
and of course as a business school
00:01:57
Professor there's also a lot of
00:01:58
interesting questions to ask that can
00:02:00
help you know businesses and companies
00:02:02
do better
00:02:03
and so now we're in getting in more and
00:02:06
more into this intersection of consumer
00:02:08
behavior and how AI is playing a role
00:02:10
how has that kind of framed your uh your
00:02:14
look at at where we stand right now you
00:02:17
know I didn't do any work related to AI
00:02:20
or technology up to about 10 years ago
00:02:23
actually my first half of my career was
00:02:26
on completely different topics and then
00:02:28
I just got interested in this because I
00:02:31
started reading about what AI was
00:02:33
becoming able to do and now going back
00:02:35
about 2014 where the first crazy things
00:02:39
started appearing like you know a a
00:02:41
agent that can understand your speech or
00:02:44
a self-driving car and I just you know I
00:02:46
have originally a background in
00:02:47
statistics and I just got curious to
00:02:49
know how this neural networks we're now
00:02:51
going to suddenly be able to do these
00:02:52
things and so it became an interest and
00:02:55
then it became a warning because then I
00:02:57
started looking more into it and
00:02:58
realized just how how powerful this
00:03:01
technology was going to be and you know
00:03:03
we're still in early days actually you
00:03:05
know and um the implication this would
00:03:08
have for my own children I were thinking
00:03:09
about what kind of education they need
00:03:11
at that time it was also the academic
00:03:13
director of a big program at uh you know
00:03:15
the university I used to work and I was
00:03:17
thinking okay what do we need to teach
00:03:19
these kids you know it's not just that
00:03:21
they would have a job next year but they
00:03:23
would have a job 10 years from now and
00:03:24
so that became it became an area concern
00:03:27
the conclusive to my professional
00:03:28
interest at that point then I thought
00:03:30
you know nobody's studying this in
00:03:31
research we don't know much about how
00:03:34
people perceive AI how I make them feel
00:03:36
and what are the barriers to adoption
00:03:38
and what kind of concerns people have
00:03:39
with it and so I basically you know
00:03:42
started developing this new line of
00:03:44
research and it's been keeping me busy I
00:03:46
would think that and you mentioned that
00:03:49
for some people uh it's somewhat of a
00:03:51
polarizing topic when you're thinking
00:03:53
about how AI is having an impact but
00:03:56
then again as you just kind of alluded
00:03:57
to the hope is that for younger
00:04:00
Generations specifically that this is
00:04:02
going to become just kind of a normal
00:04:03
part of the process of Life as we move
00:04:06
forward correct yeah and you know if
00:04:09
this is the kind of topic where it
00:04:11
leaves no one indifference I mean you
00:04:13
you I never speak to anyone who said I I
00:04:16
don't know it's not interesting it's not
00:04:17
relevant it's not important everybody
00:04:19
recognizes this a momentous change I
00:04:22
would argue historical uh change and uh
00:04:25
that you will have implication for all
00:04:27
kind of stuff in life from you know our
00:04:29
ability to have an included Society to
00:04:32
our ability to sustain Democratic
00:04:34
processes and then on the positive
00:04:36
incredible potential of improving
00:04:38
welfare well-being and the economy so
00:04:42
um I think there is another element to
00:04:44
it which is really informed by popular
00:04:46
culture we grow up with the movies and
00:04:50
you know and books and that highlight
00:04:53
just the dangers of Technology you know
00:04:55
from The Matrix Terminator and 2001 A
00:04:58
Space Odyssey or a blade done or
00:05:00
whatever it is and those kind of stories
00:05:04
do inform
00:05:06
um how we think about technology moving
00:05:08
forward and especially today that we
00:05:10
almost seem to be living in a Sci-Fi
00:05:12
movie in a way where every day we read a
00:05:14
newspaper and we just thinking wow I
00:05:16
could never thought this would happen
00:05:18
and so at that moment of upheaval and
00:05:21
change then I think some of these fears
00:05:23
are also bubbling up because of that so
00:05:25
when you think about how our identities
00:05:28
are potentially impacted by AI I would
00:05:31
imagine it's kind of a broad scope of
00:05:34
areas that you can think about at this
00:05:36
point
00:05:37
yeah you can look a lot of different
00:05:38
topics and I think a bit generally a
00:05:41
question that you can ask which I find
00:05:43
really interesting it's not just to ask
00:05:45
what do people think of AI you know a
00:05:47
lot of people want to learn about want
00:05:48
to know that and how do we you know
00:05:50
improve consumer uh belief through
00:05:53
acceptance or technology for your tech
00:05:55
company that's obviously an important
00:05:56
question but maybe maybe more important
00:05:59
or more interesting even is to think
00:06:01
about how AI
00:06:03
changes the way we think about ourselves
00:06:05
and and there is a link to to Identity
00:06:09
the identity like our human identity and
00:06:11
more specifically also our identity and
00:06:13
specific domains for example in
00:06:15
consumption or at work you know there
00:06:17
are a lot of things that we do in life
00:06:19
we don't do them only for instrumental
00:06:21
reasons to get a job done you know many
00:06:23
of them are like that you know just want
00:06:25
to get their job done but many things we
00:06:27
do them partly because that's who we are
00:06:30
you know we we uh we have hobbies we
00:06:33
have passions we have ways in which we
00:06:35
construe our personas to ourselves and
00:06:38
to other people and and those personas
00:06:41
are important and technology and
00:06:42
automation can be a threat to those
00:06:44
personas there's more and more of the
00:06:46
activities can be made by done by
00:06:49
machines so for example at the workplace
00:06:51
one big barrier to or let's say a a
00:06:56
potential stamping block for a lot of
00:06:58
tech deployments in organizations is
00:07:00
that people feel threatened by it you
00:07:02
know that they may not want to adopt
00:07:04
technology because they feel they can do
00:07:05
it better or because they're afraid and
00:07:07
now they're irrelevant so because they
00:07:09
are worried about what's coming next and
00:07:11
so some resistance is partly due to this
00:07:13
by this perceived threat that people may
00:07:15
have sometimes so communicating properly
00:07:17
about technology is important and also
00:07:20
understanding how this technology might
00:07:22
affect people's feelings of identity so
00:07:24
imagine that I am a really passionate
00:07:28
into cooking for example and there are
00:07:32
certain things that I do in cooking that
00:07:34
I don't want to be replaced by
00:07:36
maybe I'm baking bread imagine and I'm
00:07:39
okay with using a donating machine that
00:07:42
would automate the the physical labor of
00:07:44
needy the dough which is kind of you
00:07:46
know hard work and boring and but what I
00:07:49
don't want is a machine that automates
00:07:51
my cognitive skills my unique ability is
00:07:53
to understand what ingredients we need
00:07:55
and how you actually bake the bread so I
00:07:58
might be feel something like a bread
00:08:00
baking machine is be highly threatening
00:08:02
to me and you can think of that in a lot
00:08:05
of different activities maybe you can
00:08:06
think about it in your own life you know
00:08:07
what activities do you like and and
00:08:09
which activities do you feel technology
00:08:11
helps you and sometimes can hinder you
00:08:14
if you're fishing if you are you know a
00:08:16
cyclist if you are you know whatever
00:08:18
musician like it could be a lot of
00:08:19
different things is there a concern then
00:08:22
to a degree of of having that level of
00:08:25
automation of kind of a loss of skills
00:08:27
at times with human beings with some of
00:08:30
the things that you said that we'd love
00:08:31
to do that potentially we might lose
00:08:34
those at some point down the road
00:08:36
is killing and people have been
00:08:38
emphasized before and I was referring
00:08:41
more to social emotional processes of
00:08:43
feeling replaced which people may find
00:08:45
highly aversive
00:08:47
um even threatening but indeed that is
00:08:49
this issue also you know skills you use
00:08:53
it you you you develop them by using
00:08:56
them you know the the typical saying is
00:08:57
you know use it to lose it and that goes
00:09:00
also with Automation and we've had some
00:09:02
we think discussions in some
00:09:04
professional contexts for example
00:09:05
airline pilots or doctors where if you
00:09:09
don't practice certain skills you may
00:09:11
not be able to you know perform a task
00:09:14
well and especially in situations where
00:09:16
you may have a standard mode of
00:09:19
operation which is highly automated and
00:09:21
where the automation switches off when
00:09:23
there is a crisis that is a moment when
00:09:25
you really need the Newman agent to be
00:09:27
on top of things and highly skilled but
00:09:29
we've never get to practice things we
00:09:31
might not be ready when the moment comes
00:09:33
but our health our input is really
00:09:35
needed that goes also for driving a car
00:09:38
or imagine you know another pilot system
00:09:40
or would drive you everywhere and then
00:09:42
switches off when there is something
00:09:44
really difficult on the road and now you
00:09:45
haven't been doing driving for you know
00:09:47
weeks and you don't know what to do you
00:09:49
know that's an issue how does it
00:09:51
potentially impact the labor force do
00:09:53
you think and and even to a degree the
00:09:55
workplace uh that we may see in the in
00:09:58
the years ahead there's a lot of
00:10:00
discussion now in labor economics and
00:10:02
related fields on the impact of
00:10:04
automation for the demand for labor and
00:10:07
what kind of Labor so
00:10:10
um this is not my field of expertise but
00:10:12
it's a really interesting topic both
00:10:14
from a policy making point of view but
00:10:16
also just from a you know a regular
00:10:18
person point of view where you might
00:10:20
want to think about what is it that I'm
00:10:22
gonna do you know in you know five ten
00:10:24
years from now and the um consensus
00:10:28
seems to think that the current wave of
00:10:29
automation stands out in the potential
00:10:33
to automate a lot of different tasks
00:10:35
within organizations we've moved away
00:10:37
from physical labor to cognitive labor
00:10:39
and this is a big story of Industrial
00:10:41
Revolution and so we ended up from the
00:10:42
factory floor into the office right and
00:10:45
now we are being kind of kicked out of
00:10:47
the office and it's not clear with other
00:10:49
place we can go to to the song and those
00:10:52
are the fears people have but there's a
00:10:54
lot of excitement too there is you know
00:10:55
taking for example generative AI is you
00:10:58
know just a new literature springing up
00:11:00
now I was trying to understand what the
00:11:02
impact of this technology for
00:11:03
productivity image some of the early
00:11:05
data are just stunning I mean you know
00:11:06
enormous increase in productivity so
00:11:08
there's a lot to be also you know
00:11:10
excited about my own interest is a bit
00:11:13
different from this my own research is
00:11:14
not so much on demand for labor but more
00:11:16
about workers perception of uh um of
00:11:20
what it means to deploy more and more
00:11:22
technology in the workplace and just to
00:11:25
make a couple of examples we have a
00:11:27
paper came out two years ago where we
00:11:29
look at
00:11:31
um we say the psychological correlates
00:11:33
of technological unemployment meaning
00:11:37
does it feel different when you replace
00:11:39
by Machine versus when you replace by
00:11:42
another person and what we find is that
00:11:44
it does feel different and in fact it
00:11:46
feels better that you perceive
00:11:50
um replacement by a machine as being
00:11:52
less threatening generally than
00:11:54
replacement by another human worker
00:11:57
because you don't tend to compare
00:11:58
yourself to an algorithm or a robot and
00:12:01
be replaced by another person therefore
00:12:03
it's just more threatening because it's
00:12:04
not a nice comparison to make you know
00:12:06
the job went to someone else however we
00:12:09
in the same paper we find that these
00:12:11
kind of feelings are highly contingent
00:12:12
on your temporal Focus if you think
00:12:15
about how you feel what's more
00:12:18
threatening what you prefer now then
00:12:20
people do find a robotic replacement to
00:12:23
be less threatening to your sense of
00:12:25
self but if I ask you a different
00:12:26
question and now I ask you what do you
00:12:28
think about the future your economic
00:12:30
future then now the preference switches
00:12:32
around and now actually people find
00:12:34
robotic replacement more threatening you
00:12:37
know robotic replacement is a cue for a
00:12:39
skill obsolescence you know we're
00:12:41
talking about risk healing and
00:12:42
riskilling so the moment that I get
00:12:45
fired and I get replaced by another
00:12:47
marketing Professor let's say I can go
00:12:50
out there and try to look for another
00:12:52
job like the one that I lost but if I
00:12:54
get replaced by an algorithm which that
00:12:56
becomes an early Professor well you know
00:12:58
they're never gonna be another job for
00:13:00
me so that's that's basically the logic
00:13:03
of this temporal switch in another line
00:13:05
of research this is a new one we haven't
00:13:07
even published this paper uh what we're
00:13:08
looking at is um uh people with the
00:13:12
stereotypes about Ai and how they can or
00:13:15
beliefs about Ai and how this can spill
00:13:17
over to uh to the people who somehow
00:13:21
were connected to the eye so imagine
00:13:24
um a company relying on AI for employee
00:13:28
selection that's increasingly common
00:13:30
nowadays and what we're finding is that
00:13:32
people hold stereotypes about the people
00:13:34
that get recruited through AI systems
00:13:36
such that if you're an employee and
00:13:38
someone a colleague of yours for example
00:13:40
learned that you were hired through an
00:13:43
AI driven selection process people tend
00:13:45
to hold the belief that you have
00:13:47
inferior interpersonal skills
00:13:49
they also hold the belief that you have
00:13:52
Superior analytical skills so they kind
00:13:54
of reflect the stereotypes that we have
00:13:56
about technology onto the people that
00:13:59
were selected by technology and that's
00:14:00
probably actually in some early data we
00:14:03
have and other data that doesn't just be
00:14:05
the case actually AI can be very good at
00:14:07
selecting based on interpersonal skills
00:14:09
but people don't always believe
00:14:12
does the impact of chat GPT play a role
00:14:16
into that thought process of how AI as a
00:14:20
replacement can be sometimes less
00:14:22
threatening I mean chat GPT I think for
00:14:25
a lot of people is believed to be a
00:14:27
benefit at least in the early going of
00:14:29
the work that it can do
00:14:31
not so much as a replacement but as a as
00:14:34
a support for what somebody is doing in
00:14:36
the workplace
00:14:39
its launched in November has been I
00:14:43
think a very important moment in the
00:14:47
diffusion of AI technology in society I
00:14:50
mean this technology of generative AI
00:14:53
has been around already for a little
00:14:55
while but before it was confined in
00:14:59
relatively narrow uh you know coordinate
00:15:02
of society whether that was a tech
00:15:03
sector or Academia or some other place
00:15:06
like that and the deployment at scale of
00:15:10
a technology like that where in order to
00:15:13
now being able to use a generative AI
00:15:16
algorithm like cha GPT you would only
00:15:18
need an email address semester three
00:15:20
minute registration process makes
00:15:22
basically adoption possible for
00:15:24
everybody and you know people have been
00:15:26
so intrigued by this is this you know
00:15:28
probably you know the listeners I've
00:15:31
heard this already but it has been it's
00:15:33
a product with the fastest adoption ever
00:15:35
recorded it took only five days to reach
00:15:37
a million users it took less than eight
00:15:40
weeks to reach a mere 100 million users
00:15:42
and and so basically everybody's trying
00:15:45
it and playing with it and the first
00:15:47
experience is just amazement you know
00:15:49
that an algorithm can do this and I
00:15:52
believe I probably speak for many of us
00:15:54
but it's not obvious that any of us who
00:15:56
think something like that we would be
00:15:59
able to see in our own lifetimes and
00:16:01
it's happening so fast so
00:16:04
um yeah this is the exciting around it
00:16:05
and of course most people that make an
00:16:07
account for jet GPT what do they do with
00:16:08
it silly things you know you want to
00:16:10
write I don't know you ask him to write
00:16:13
your own bio and just to see what he
00:16:14
does or you ask it to write a rap song
00:16:17
about you know Shakespeare or some kind
00:16:20
of silly thing a gimmicky thing but it's
00:16:22
very easy to understand just how
00:16:24
pervasive the impact of this technology
00:16:26
can be in a lot of different jobs I mean
00:16:28
you can think about your own job what do
00:16:30
you do in a day and however many of the
00:16:32
things that you do in a day could be
00:16:34
done at a reasonable level by technology
00:16:36
like this with fairly limited prompting
00:16:37
and so of course it's not as good as a
00:16:40
lot of the stuff that we do but it's
00:16:41
good enough for a lot of users and maybe
00:16:43
in some context it's actually as good or
00:16:45
maybe even better than what we can do
00:16:47
ourselves well let me let me ask you
00:16:49
then also in in your Realm of the
00:16:52
marketing world of how AI you think will
00:16:56
be continued to play a role moving
00:16:58
forward in terms of its connection with
00:17:00
the consumer and what the consumer will
00:17:03
be expecting back from companies who are
00:17:05
obviously using this technology
00:17:07
yeah definitely consumer expectations
00:17:09
are going up very quickly so now for
00:17:12
example think about chatbots your
00:17:14
expectation about the chatbot was going
00:17:16
to be able to do it's gone up a lot in
00:17:18
the last year or two and maybe from
00:17:19
expecting a very rudimentary and
00:17:21
probably you know unsuccessful kind of
00:17:23
interaction you probably now have
00:17:25
expectation of interaction can be pretty
00:17:27
smooth and probably affected so I think
00:17:29
that means that companies cannot rest on
00:17:31
their Laurel that they keep improving
00:17:32
they need to deploy technology for the
00:17:34
benefit of customers and when they don't
00:17:36
do that they'll likely be left behind
00:17:39
but there are also issues around AI
00:17:41
safety for example I know it's that the
00:17:44
technology gets deployed at scale in
00:17:47
society and there's more and more people
00:17:48
are interacting with it there are also
00:17:51
situations where we can see dangers for
00:17:53
consumers and for example you know with
00:17:56
already media stories about people
00:17:58
falling in love with AI or you know
00:18:00
living the the husband because AI advise
00:18:02
them too and it was this famous story in
00:18:04
the New York Times about being AI a
00:18:07
couple of months ago
00:18:08
you know the journalist was that was
00:18:10
back to to leave the uh the wife or for
00:18:13
a child of being Ai and uh and so
00:18:16
clearly this is potentially an issue
00:18:18
especially if you have consumers
00:18:19
potential with mental health issues who
00:18:21
interact with technology you could
00:18:23
easily see how this can become
00:18:24
problematic the thing to understand here
00:18:26
that these algorithms are are machine
00:18:29
learning algorithms and make predictions
00:18:31
and respond in contexts you cannot know
00:18:33
beforehand what they're going to say so
00:18:35
if a consumer for example expresses a
00:18:38
wish to end their own life it's not
00:18:41
clear that this algorithm will say
00:18:43
anything helpful or or safe and the the
00:18:47
person who called this technology also
00:18:49
cannot know it's not like a scripted
00:18:50
where you know someone with clinical
00:18:53
understanding of that circumstance will
00:18:55
be able to you know embed safe
00:18:58
instructions in the in the system the
00:19:00
system does what the system does and and
00:19:03
might not be what we needed to do so
00:19:05
there are issues around AI safety really
00:19:08
we're going to see a lot of actions also
00:19:10
in the regulation side but I think it's
00:19:14
you know exciting and also scared at the
00:19:16
same time yes so is it kind of important
00:19:19
then to kind of frame the use of AI as
00:19:24
complementary to what we already have in
00:19:26
society instead of relying upon it as
00:19:29
kind of the the next step and how to do
00:19:32
everything in in our lives yeah I agree
00:19:35
Dan I think you make a good point and I
00:19:36
think too much of the discussion around
00:19:39
the AI over the last few years has been
00:19:42
of the kind I call human or AI human or
00:19:46
machine meaning we've been thinking
00:19:48
about how we can take the human out of
00:19:51
the equation take for example
00:19:52
self-driving car the way that the
00:19:54
algorithm been working is that you have
00:19:56
a human driver driving on the road with
00:19:58
an algorithm recording everything that's
00:20:00
happening and then the algorithm would
00:20:01
learn what to do when a particular set
00:20:04
of parameters face the algorithm so now
00:20:07
in this situation what would a human do
00:20:09
copying the human this is basically how
00:20:11
a lot of this machine learning algorithm
00:20:13
work but I think we need to change gear
00:20:16
now and I think we need to go into the
00:20:18
mindset I call human and machine not
00:20:20
just understand how we can mimic a human
00:20:23
decision process so we can maybe do it
00:20:25
instead of the human but focus more on
00:20:28
how do we Leverage The Unique
00:20:30
capabilities of algorithms and of humans
00:20:32
which are different and complementary in
00:20:34
order to improve business processes so
00:20:36
it's not about you know replacing the
00:20:38
human I think it's making the human more
00:20:40
effective more inspiring more productive
00:20:44
thank you for listening to the ripple
00:20:46
effect we hope you found this episode
00:20:47
informative and engaging don't forget to
00:20:50
subscribe and leave us a review so that
00:20:52
we can continue to bring you the best
00:20:54
Insight from the Wharton School
