00:00:00
So in general, if we're talking about very large firms, particularly firms that have experienced corruption in ways
00:00:07
that can be captured, right, >> AI is great. When we're talking small to
00:00:12
medium firms, don't generate a lot of data, AI can be counterproductive. It can it can yield hallucinatory responses
00:00:21
that that could hurt individual people or hurt transactions or relationships. Welcome to the Ripple Effect, the
00:00:28
podcast that takes you on a journey through the minds of Wharton faculty. I'm your host, Dan Looney, and in each
00:00:34
episode, we'll be diving deep into the inspiration behind the groundbreaking
00:00:38
research that Wharton professors have conducted and exploring how their findings resonate with the world today.
00:00:45
Well, as we continue to hear that artificial intelligence is the technology that every firm should
00:00:52
implement, [music] our guest today questions whether that can be the case when you're thinking about things like
00:00:57
anti-corruption efforts by countries and agencies around the globe. Pleasure to be joined here in studio by Philip
00:01:04
Nichols, professor of legal studies and business ethics here at the Wharton School. He wrote about that question
00:01:08
exactly in a recent article for the American Business Law Journal. Phil, great to see you. How are you?
00:01:15
>> Fabulous, Dan. It's always good to see you. >> Good to see you as well. So, I I guess
00:01:19
let's dig in right into the guts of this because everybody thinks that every
00:01:23
company needs to have AI in everything that they are doing. >> Exactly. >> How does potentially AI fit in or maybe
00:01:30
not fit in in and around the efforts to prevent bribery, corruption, etc.? >> Great question. Um to answer that
00:01:40
question we need to understand two things about corruption and two things about artificial intelligence.
00:01:49
The first about corruption it manifests itself differently everywhere. The corruption you
00:01:56
experience in one country is different than the other in one industry different than the other in one firm different
00:02:03
than the other. And that means data is not fungeible. data is not easily translatable or usable
00:02:10
>> right >> from one to the other. >> The second thing to understand about
00:02:14
corruption and and in compliance in general is that misbehavior corruption uh occurs in the shadows.
00:02:22
People don't want to generate a lot of data about it, right? Two things to understand about um artificial
00:02:29
intelligence in addition to it uses a lot of energy and it requires a lot of processing power,
00:02:34
>> right? It requires a a solid model of what the world looks like and it requires a lot of data,
00:02:43
>> right? >> Therefore, if we put those two things together and and not even talk in
00:02:49
specific about what we're asking artificial intelligence to do, but just whether artificial intelligence is
00:02:55
viable with this particular realm of compliance, >> right? >> We don't have the data and we don't have
00:03:01
the model. And therefore what we're likely to get when we ask questions of artificial intelligence is
00:03:08
hallucinations or nonsense. >> Right? And I guess to a degree with the issue of compliance
00:03:14
>> then you have to bring in the point that the rules in Europe are different than
00:03:20
what we see here in the United States and [snorts] other parts of the world. And so
00:03:24
>> we think about the use of AI as something that is global. We think about
00:03:29
the ways of protecting from corruption as global. Yeah. >> Yet you have these different sets of
00:03:33
rules in place. >> Yeah. The European Union's rules regarding artificial intelligence really
00:03:39
protect individual dignity >> whereas the rules for the use of AI in North America, particularly United
00:03:47
States are are much looser, >> right? And so using AI to like investigate one of your employees, are
00:03:56
they accumulating wealth surreptitiously, >> right? >> You wouldn't be able to do that in
00:04:00
Europe. And if you're a transnational firm, you really need to comply with Europe's they're they're kind of like
00:04:07
the gold standard of use of AI. And so there's a lot of things like you point
00:04:12
out that we won't be able to like dig into, investigate, etc. because that really would be an infringement on the
00:04:21
dignity of the people who work with us and for us. >> So a lot of what is also asked involves
00:04:28
up-to-date process, up-to-date software, etc. >> Yeah, >> at least right now it seems like up
00:04:34
toate in the realm of what we're talking about is AI. It's it's taking that next
00:04:40
step. >> That's my concern. I I you know someday I I have no doubts that someday new
00:04:46
forms of artificial intelligence will be very useful empathetic or empathic AI intuitive AI but right now most of our a
00:04:56
either is either in the form of large language models or is it is self-learning and therefore can
00:05:02
categorize and make a sense of large ma large masses of unsorted data >> right
00:05:10
>> neither one of those right now fits the the the things we need to do with
00:05:15
corruption. And so the notion that oh we adopt artificial intelligence and therefore we're up to date and doing the
00:05:23
best we can >> really doesn't work when we get into the weeds and think we want to solve these
00:05:30
problems not just put a little sticker on that says hey we're using AI >> but are there processes components of
00:05:38
what is being done now to try and prevent corruption prevent bribery etc >> where you see AI could have absolutely
00:05:46
>> a level of impact. >> Absolutely. Two areas. One is areas um is in generally warning people. So we
00:05:54
call that red flags. So sorting data. Yeah, your firm does this this does this transaction, this
00:06:03
action, whatever a particular way for thousands of times and then there's one
00:06:09
that's different, right? >> Well, that's a red flag. doesn't mean
00:06:11
it's necessarily a misbehavior, corrupt. It just means that someone should take a
00:06:17
look at it and see why it's different. AI is great for that. And the other way
00:06:23
that AI is really useful is when we do have huge masses of data, right? >> So the Panama papers, the Pandora leaks,
00:06:32
etc. or an individual firm, Seammens, Walmart, right? These are are are corruption incidents of corruption that
00:06:44
did yield massive amounts of data and there artificial intelligence can be useful. So in general if we're talking
00:06:53
about very large firms particularly firms that have experienced corruption in ways that can be captured right
00:07:00
>> AI is great. When we're talking small to medium firms don't generate a lot of
00:07:06
data, AI can be counterproductive. It can it can yield hallucinatory responses that that could hurt individual people
00:07:15
or hurt transactions or relationships. >> Is there a level of this? And when you
00:07:20
think about the rules that have been put in via Europe around GDPR in the last decade or two?
00:07:25
>> Yeah. >> And they were seen to be at the forefront of the move to try and protect
00:07:30
personal data. Yeah. Is there a way that you could say a region of the world because they have kind of taken the
00:07:37
steps forward maybe finding a path sooner rather than later to implement AI to be able to use
00:07:43
this in the process or are we still far away from that at this point? >> Um I think we're far away from it in
00:07:49
general. Right. >> I your question is a really interesting question and and it's a question that
00:07:55
that goes beyond just compliance or you know detection of misbehavior. the Europe, China, and the US all have
00:08:04
very different approaches, >> right? >> Which one of those approaches is going
00:08:08
to yield the most accurate detection of misbehavior? I I think there's an argument to be made for Europe's
00:08:16
approach because un unlike you [snorts] know our use of for example um AI and hiring right now Europe's approach might
00:08:26
lead to a better model which is half of what makes AI work model and data >> and Europe might lead to a better model
00:08:37
that I mean it's a really interesting question that we don't really know
00:08:42
enough about I mean we're what just a few years into this transition. >> Yeah.
00:08:47
>> But there there's a a good argument for Europe. There's also a good argument for
00:08:51
the US in that >> you know just letting people run wild and do whatever they want to with AI,
00:08:57
you know, might yield the magic bullet too. >> So then from the same perspective then
00:09:02
when you think about how regulation plays a role in how companies operate. >> Yeah. I guess to a degree the unknown of
00:09:09
AI that we still have in a lot of areas really still puts us in an unknown in terms of how it would impact with
00:09:17
regulation as well. >> Yeah. I I the the absolutely the the thing that I'm most concerned
00:09:25
the thing I wanted to experiment with and play with was since AI seems to be the magic band-aid for everything in the
00:09:32
world and you're not up to you know you've got to have AI AI or you're
00:09:37
behind is that true right now and the answer I came up with is no >> there's plenty of areas where AI is
00:09:46
neither not helpful or detrimental and compliance regulation is is one of them. >> Is that going to end up then leading us
00:09:55
to kind of rewriting rewriting policy and regulation in some of these areas if we do get to a point of of implementing
00:10:02
AI? >> Yeah, I I imagine just like with the industrial revolution um rules changed because the world we
00:10:11
work in and the world we live in has changed, >> right? in that in in a number of years
00:10:17
the regulations the the rules for business behavior will be written with an eye toward the fact that most
00:10:25
businesses incorporate to some extent varying degrees um because by then we'll
00:10:31
have a more realistic understanding artificial intelligence >> so in in the times that you and I have
00:10:36
talked about these issues yeah >> I think it's fair to say when you think
00:10:39
about corruption you're thinking about it at the business level at the firm I
00:10:44
do. Absolutely. I'm here at Wharton. >> When you're when you're thinking about
00:10:48
bribery, >> it's more of a personal, you know, one person to one person, two people to two
00:10:53
people connection. So, how does that difference potentially impact the thought process of using AI to kind of
00:11:02
mitigate the issues around bribery? >> Yeah. Um so if we go to the other end
00:11:07
and talk about you the use of artificial intelligence in government in businessto
00:11:13
business transactions >> um there is great potential there's great potential for abuse
00:11:19
>> right >> there's great potential particularly since AI can leverage at huge scales
00:11:24
>> right >> for all kinds of misconduct but there's also great opportunity for reducing or
00:11:32
almost to nothing that kind of human interference and discretion that that can lead to corruption and the digital
00:11:41
platform infrastructure in India and Aadhar in India are great examples of technology I mean and here we're not
00:11:50
even talking self-learning or what you know whatever the AI bucket is >> right
00:11:54
>> but where technology has done exactly what you describe and significantly
00:12:00
reduced low-level corruption significant significantly redu increased the amount
00:12:07
of money that people are entitled to from various government distributions that they're getting at the village
00:12:14
level. So at the other end, the not not the compliance end, but the the government operation and regulatory end,
00:12:21
yeah, there's potential for abuse, but there's also great potential for cleaning up corruption, which is a
00:12:27
really neat story. What would then some of the policies or rules you think that would have to be put in place in order
00:12:35
to have [clears throat] that kind of effectiveness? >> Yeah, one is giving people digital IDs.
00:12:41
Uh right, you know, we kind of juryrigg that in the United States and in North America right now. We like use
00:12:48
somebody's cell phone number >> um the email address. But the if you look at places like Estonia where they
00:12:56
they they just from the ground up once they achieved reach achieved independence
00:13:02
um digital IDs that opened up so much for working with technology which event you
00:13:10
know will include of course self-learning and and artificial intelligence. You look at Adhar in India
00:13:19
once you have that digital ID even at the village level it opened up access to the kinds of things that you're talking
00:13:25
about. So that's one one thing that you know policy countries are leapfrogging those of us
00:13:34
who still use what's your mobile phone number. >> Yeah. >> Um that's just a really simple kind of
00:13:39
thing. Now how one approaches the mass leveraging the the depersonalization all of these things
00:13:49
with with still some kind of form of justice and and and you know fairness that's something that you know the in
00:13:59
the industrial revolution it took them 80 years at least to figure out. hopefully will be a lot faster, but we
00:14:08
kind of need to see what intuitive AI looks like, what empathic AI looks like before we start developing those kinds
00:14:17
of regulations. And maybe I'll throw in the cynic in me a little bit, but don't
00:14:20
you don't you also have to assume that if we are looking at ways to potentially
00:14:26
implement AI to prevent this type of criminal activity, Yeah. The criminals are out there looking for ways to use
00:14:32
the AIB to conclude this activity. >> Absolutely. Without question. And and
00:14:37
and and it does give them huge potential. Um the leveraging gives them huge potential from misconduct.
00:14:46
The the ability to crack other, you know, previously secure forms of communication or interaction, huge
00:14:54
potential. So there's no question. It's not that AI eventually it's it doesn't work now,
00:15:02
>> right? >> But even in the future, it's not a magic bullet. It's a tool that we need to use
00:15:06
wisely. >> And and we can never think of it as just well, we're done now. AI is taking care
00:15:12
of everything. It's >> it it may become sentient. It may become whatever, whatever, whatever, but it's
00:15:18
still a tool, not a magic bullet. And I'll finish on this because even with the implementation of AI in so many
00:15:24
firms right now, the expectation is you still have to have the human component in there as part of it and I think you
00:15:30
just kind of alluded to it. Yeah. That even if you're using AI in terms of trying to prevent this level of criminal
00:15:36
activity, you still need to have the human component in there monitoring and helping go through the process.
00:15:42
>> Absolutely, without question. >> Phil, great to see you again. Thanks
00:15:46
very much. >> Dan, always excellent to see you. Thank you. >> Thank you. Philip Nichols, uh, professor
00:15:49
of legal studies and business ethics here at the Wharton School. Thank you for listening to The Ripple Effect. We
00:15:55
hope you found this episode informative and engaging. Don't forget to subscribe
00:15:59
and leave us a review so that we can continue to bring you the best insight from the [music] Wharton School.