00:00:00
Well, as we are in the middle of graduation season, everybody thinking about that next job,
00:00:06
and one of the components that they will be dealing with as we move forward is the impact
00:00:11
that artificial intelligence is having and will have down the road. Pleasure to be joined by
00:00:16
Peter Capelli, Management Professor here at the Wharton School. Peter, great to talk again. How
00:00:20
are you, sir? I'm good. Dan, how are you? I'm doing well. Let me start with this time of the year,
00:00:26
and you having been at Wharton so long, in the conversations you have with students, what is
00:00:31
kind of the expectation you give of them of what they're going into? Well, I mean, this won't surprise anybody. There's just tons of uncertainty, right? And
00:00:45
the reality of AI is lagging, or perception is lagging the reality by a lot. So I've been
00:00:54
spending a fair bit of time just trying to talk to organizations that have actually introduced AI,
00:01:00
as opposed to people who are just talking about what it will do. Right. And, you know, the gap is
00:01:05
just enormous. I mean, one of the things that you learn from even looking from survey data is
00:01:11
there's not nearly as much effort to really introduce AI into the work stream of organizations
00:01:18
as you would think. That it's mainly people using it at their desk to draft better memos and that
00:01:24
sort of stuff. But they're not taking over tasks, except for that simple stuff that you would do,
00:01:30
like, you know, write letters or answer email or something, or summarize documents, right?
00:01:34
Yeah. We've seen a lot of that. Yeah. But really putting it into the workplace
00:01:38
where it would take over tasks, we're not seeing that. The evidence about it taking over jobs,
00:01:43
there's really, like, no good evidence of that. And lots of evidence that it's just not happening.
00:01:49
Here's the latest one I just saw. And that is a survey of chief financial officers who you would
00:01:55
think might have an incentive to exaggerate this because Wall Street likes the idea that you would.
00:02:00
Right. Right. And they report that AI has reduced jobs by 0.4%. So four-tenths of a percent.
00:02:10
Yeah. Much. And unfortunately, it's become the story that is kind of the hot-button topic out there.
00:02:18
And so when you hear a company cut jobs and there is something attached to it saying,
00:02:25
well, it's because of the advent of AI, how do you respond to that? Well, there have been sort of three stages of the explanations, right? And I don't think any
00:02:36
of them have been true, to be honest. The first was, you know, that AI has cut, is reducing our
00:02:44
headcount. And if you look at that, you discover, and if you ask, you know, has it actually taken
00:02:49
over jobs? The answer there has been no. Right. And then the second one was, Amazon in particular
00:02:55
was very upfront about this with its 14,000 layoffs. We anticipate that it will take over
00:03:02
jobs, which if you think about it is kind of a very bizarre thing that you would cut jobs now
00:03:09
in anticipation that eventually they'll be taken over. So where's the work going to get done?
00:03:14
Right. Right. That didn't sound right either. The thing we have heard now is a third explanation. And that is
00:03:21
that the layoffs are because we need to save money to invest in AI. Right. Yeah.
00:03:29
Mark Zuckerberg said this about Meta the other day, quite literally, and you're hearing it
00:03:34
elsewhere as well. So I did a little look into that and whether that sounds credible
00:03:40
and just run it through for you a little bit. Most of the big tech companies that are making
00:03:45
these kinds of arguments, they've got over a hundred billion dollars in cash or so. They're
00:03:50
cash generators, right? Yeah. And they've got market cap of three, $4 trillion. I mean, they're enormously valuable companies.
00:04:01
And they're talking about making big investments, over a hundred billion dollars
00:04:06
of investments in AI. If you ask the investment community and the analysts, where's the money
00:04:12
coming from? They say it's going to come from their revenues, which are at record levels.
00:04:17
And then if you look at what you're going to save from these layoffs, some estimates,
00:04:23
you lay off 10,000 managers, you might save $2 billion. $1 billion has got to go to basically
00:04:28
help them leave the cost of severance and things. So upfront, you're saving a billion dollars
00:04:34
on an investment of a hundred billion, where you've already got a hundred billion dollars
00:04:39
in cash sitting around. Come on. It's not that big a thing. It's just a drop in the bucket.
00:04:45
And the only reason it is happening is if you believe you really didn't need these people in
00:04:49
the first place, right? Right. And I think that's the reality. And why that is, is a much more complicated question.
00:04:56
So then, is some of this then potentially still a little bit of a leftover from the pandemic time
00:05:03
when we saw companies adding bodies and adding bodies and adding bodies, and they kept people
00:05:11
on to a certain point? And maybe now is this point where they're starting to trim a little bit?
00:05:16
Ooh, that's a really good point. And I just was trying to float that argument to some people. I
00:05:21
was at a conference of HR people yesterday, and we were trying to figure out why are these cuts
00:05:30
happening now? Because business is booming, right? I mean, one of these companies are arguing we're
00:05:34
in any financial trouble. And there was an undercurrent of kind of belief that what was
00:05:40
going on here was kind of payback. Because it's not just that we're getting layoffs. It's the
00:05:47
pushing people to come back to the office. It's squeezing budgets. You may have seen this story
00:05:52
about Deloitte and Zillow and some other companies cutting paid time off days. Yeah.
00:05:58
In the middle of, you know, what is a business boom? And so, you know, why is that? And that
00:06:04
sounds like a credible explanation, at least to a lot of HR people. But going back to what you were talking about before about AI, it's obvious from so many people
00:06:15
using AI in their regular life, the benefit that it can provide to a lot of those, as you said,
00:06:21
smaller tasks to be able to make sure that you're, you know, you're able to get work done faster and
00:06:28
cleaner. And that's going to be the benefit for companies moving forward in that their, you know,
00:06:34
profitability is going to rise over the course of time because things are moving smoother.
00:06:40
Yeah. I think to the extent to which it raises, you know, productivity and saves you some time,
00:06:47
which translates into productivity increase, this is exactly what an economy wants to happen,
00:06:53
right? We've gone a very long time with pretty low increases in national productivity. And this
00:06:57
is exactly what we would like to see. What I can tell you, we're looking at these companies that
00:07:03
introduce it, though, is that it takes a lot of effort to put these tools into the workflow. It's
00:07:10
a lot of organization change stuff. It requires a lot of help from your employees. You know,
00:07:16
even in white collar jobs, the idea that you're just going to pop an AI agent into it, or you can
00:07:22
sprinkle it like pixie dust over the organization, you know, you got to figure out where to do it.
00:07:27
You've got to get those employees to help you train even a chat GBT like instrument so that
00:07:34
it's giving you the answers that really are useful. You got to check it to make sure it's not giving
00:07:39
you stuff that's dumb. Every time you change the question a little bit, you got to be checking that
00:07:44
stuff. So you do need employees help on this. And if you're not getting it, it's not going to work
00:07:51
very well. What does that mean then for the younger generation coming into the workforce?
00:07:58
I think what you're basically saying is if you're more AI proficient, the better off you're going to
00:08:04
be as you move forward here. Well, you know, I think there's a couple of myths about that. The
00:08:10
first is in none of these places that I see that there was any skill involved in using AI tools.
00:08:17
Right. Right. I mean, the reason that they're so prolific is they're so easy to use.
00:08:21
What you do have a shortfall in is on the management side. This is really organization
00:08:26
change stuff. And one of the things we've heard on the tech consulting side is their need and their
00:08:33
willingness now to bring in management consultant help to try to get them to figure out how to make
00:08:39
this stuff work. You know, there's another thing that also I'm seeing fairly often now, and that is
00:08:46
that, you know, this notion that the simple jobs will all go away because AI can take them over
00:08:52
is not looking true. And the reason for that is because to take over simple jobs, the first point
00:09:02
is it doesn't save you much money. A simple job don't pay much. So if you eliminate a few minimum
00:09:08
wage workers, you're not saving a lot of money. And often the tools that you need to take over
00:09:15
simple jobs, which are kind of yes, no jobs, you know, is this an application? Is this an appeal?
00:09:20
You know, stack them, that sort of stuff that requires machine learning tools. And those are
00:09:26
quite expensive to build. And if you're the CFOs, we're starting to hear CFOs now no longer writing
00:09:33
a blank check for AI stuff on the hope that it will cut headcount. Right. And so now they're saying,
00:09:40
OK, show us what the return is really going to be here. The complication there is the costs are all
00:09:46
up front and the benefits come later on with increased productivity. But that might be years
00:09:52
down the road. And it's really a burden to get an ROI that's going to be reasonable if you're front
00:09:59
loading cause benefits come years later. Especially if you're talking about the difference in years and
00:10:04
seeing the benefit being 10 or 20 years down the road to take that hit at the front end of your
00:10:10
statement and not see it for such a long period of time. Oh, yeah. I mean, I mean, I think in this
00:10:15
case, you know, you might see it in two, three years, something like that. Right. Start to see it.
00:10:19
But, you know, the upfront costs can be really big, especially if you're including the costs of
00:10:24
organizational change and the time and effort to try to get these things to work. It's it's, you
00:10:30
know, sometimes really hard. So I think the idea that all simple jobs will be automated, it's not
00:10:35
looking like that. In fact, the big opportunity is really in white collar work. It's not so much in
00:10:41
frontline work. Right. And I think the assumption was and I'll finish up on this is that this was
00:10:47
going to have a greater impact on blue collar jobs, especially ones that had that connection
00:10:53
with the public. But for many companies, they want that still they need to have that connection with
00:10:58
the public. Having that human being, you know, conversation, it's valuable or else they're going
00:11:04
to lose customers. Yeah. Wait staff are not going away. Right. And store owners, you know, they
00:11:10
still need clerks and those sorts of jobs. I think one of the reasons there is so much panic about
00:11:15
this stuff is precisely because it was it had the possibility to affect middle class jobs and
00:11:23
college graduate jobs. And I think what what we are seeing is the slow adjustment of employers
00:11:28
to realize that, OK, maybe I like in consulting or accounting can take over the tasks that we
00:11:37
initially had college graduates do. And so and at first they said, OK, well, we don't need these
00:11:44
people. And then somebody maybe smarter thought, OK, if you don't hire people this year, you won't
00:11:49
have any two year veterans next year. Right. And why don't we just give them more to do now?
00:11:56
So I can do a lot of these things. All right. Have them do the tasks, manage those. Let's
00:12:01
give them some smarter things to do. Right. And that's what we're seeing. The consulting
00:12:05
companies have already come around to doing. Peter, always great to talk with you and get
00:12:10
your insight. Thank you, sir. Thank you, Dan. Nice to see you. Always great to see you as well.
00:12:14
Peter Capelli, management professor here at the Wharton School.