00:00:00
There's been a discussion for quite some
time that artificial intelligence was
00:00:05
going to have an impact, potentially
negative, on the workforce. Especially when you're talking about jobs
like blue-collar jobs.
00:00:14
We could see robots
replacing humans in many of the tasks that require
some level of labor.
00:00:21
But there is now more and more thought
that there will be somewhat of a blue-collar transition or revolution in
terms of the types of jobs that humans
00:00:33
will still need to be called on for in
this new AI world. Pleasure to be joined by Lynn Wu,
who's an Associate Professor of Operations,
00:00:41
Information and Decisions here at the
Wharton School. Lynne, great to catch up again.
How are you?
00:00:47
I'm good, thank you for having me. Thank you.
All right, so I'm one of those people.
00:00:50
I've heard all of this conversation
going on. There's obviously been a level of concern.
00:00:56
Where should we believe this? Because seemingly this idea of a
blue-collar revolution or readjustment or
00:01:04
however you want to— is coming into play,
isn't it? It is. I think it's an exciting time,
because for the longest
00:01:12
time we worry about blue-collar
workers, and we still do. And now with the general rise of this type
of super incredibly capable artificial
00:01:24
intelligence tools, it can enable a lot of
blue-collar work as well. What are the types of jobs then,
if humans are going to be replaced by
00:01:34
robots in some areas,
what are the types of jobs that humans will still
need to be counted on for?
00:01:40
Absolutely, that's a great question. So let me just be clear. A lot of the digital AI tools we use today
are in the virtual world.
00:01:48
They're not co-existing, occupying the
same space and time as we do. When that happens, when robots have the
type of intelligence accelerating the way
00:02:00
that digital tools in our computers,
in our laptops, in our phones are doing, then there's a different story.
00:02:07
For currently, there's a lag between the
physical AI and the digital AI. So that means in short terms, medium
terms, that physical manipulation in the
00:02:18
unstructured environment. Like, you know, people who do the last mile of
delivering your product to your door,
00:02:28
delivery workers, or healthcare workers
who lift you out of the bed, doing services for you when you're bedridden,
right?
00:02:35
These kinds of work are still going to be
very much in demand. And furthermore, the
type of job that we care
00:02:42
about is people who can
use AI really effectively. I mean, it's how do I use AI to solve
really difficult problems?
00:02:50
And problems with, you know— under
certainty and easily verified, that, AI can do already.
00:02:57
But problem solving
under very high uncertainty and verification can
take a long period of time.
00:03:03
That requires human judgment and
responsibility. And that's where humans can shine.
00:03:07
And lastly, you know,
interpersonal social intelligence is not something
AI can ever replace.
00:03:12
- Right.
- At least yet. - Right. And do you expect that even with this shift you talk
about in the medium term,
00:03:19
are these changes that are going to be
going on, are they going to be more long term or is
there kind of a shelf life to them?
00:03:28
There might be a shelf life to them just
because of scaling law, in that the ability
00:03:32
of artificial intelligence has been grown
exponentially. Like in 2022, when ChatGPT
first released, we didn't think
00:03:40
it could do the coding kind of
task that we're seeing today. So we generally tend to under— we
generally do not understand exponential as
00:03:51
a human brain because we are linear
thinkers. We're not designed to think exponentially.
00:03:56
So I would say in the long term,
this could continue. And then robotics,
physical AI, will be entering
00:04:04
a scaling law just as
we see with digital AI. And then all of this will be rethought
again.
00:04:10
Right. But right now, at least in
the medium five, ten years, we're not seeing that yet because
it is a difficult problem to solve.
00:04:17
Now, you mentioned healthcare a moment
ago. I would think there are certain
sectors where there's almost
00:04:24
kind of a platform to how much
we can expect AI to contribute. Like in healthcare, I don't think there's
much doubt that doctors and nurses,
00:04:33
the physical need
for those jobs is still going to be there for
a long period of time.
00:04:40
Absolutely. I think healthcare is a good example,
and this applies to all jobs. We have to rethink about what jobs are
actually encompassing.
00:04:50
What is it encompassing? So the doctor's jobs have already changed. If you think about
radiologists, right, they don't
00:04:57
spend time reading,
detecting cancers anymore. But they spend a lot of time communicating
the findings to the primary care or
00:05:03
whatever the cancer doctors, to formulate a
diagnosis or treatment plan. And their jobs are shifting in the sense
the job they did before may not be the job
00:05:14
they're doing right now, but
the radiology job will always be there, just like most of
the jobs classes we see today
00:05:21
will be there, just what we're doing
within that job will be changed. Does this then change the thought process,
and this is maybe a larger scale question
00:05:30
about AI and the workforce moving forward,
of how the younger generation may need to
00:05:36
think about what their career might want
to be 10, 20, 30 years down the road? I think it's never too
early to think about what
00:05:45
you want to do with life,
and it's okay to change. And I think right now for
young people, this is a really
00:05:52
exciting time because the
rule book was thrown out, right? And you got to redesign your own rule
book.
00:05:59
This generation of people gets to redesign
it. They don't have to play
the rules that we set up
00:06:04
50, 100 years ago since
the Industrial Revolution. So it's a really
exciting time for you to
00:06:08
control what you want
to do with your career. And how do you do that? You got to use AI.
00:06:13
You got to use AI, daily life. Use it, how
to do what you want to do more effectively
00:06:19
and more efficiently. And that's where you're going to have the
biggest edge.
00:06:22
And I think it's interesting also,
I saw a study the other day, but it seems
00:06:26
like there's already a recognition by
younger generations about some of this.
00:06:31
Because the report said that you've got
about 13 or 14% of undergrad students who
00:06:38
are already rethinking their major because
of artificial intelligence. So the younger
generation we know is more
00:06:44
tech savvy than some
of us older individuals. But it truly is an understanding,
it seems like already, that you need to
00:06:53
really look at what you want to do because
of the impact that AI is going to have.
00:06:58
Exactly. So if I were a young 20-something,
I absolutely would seize the opportunity.
00:07:02
I will use AI and see whatever I can do
with it. You know, redesign your rulebook.
00:07:08
You get to redefine what's going on next
50-100 years. That's a really exciting time for this
generation.
00:07:15
Overall then, how do
you think the workforce is going to be impacted as
we move forward here?
00:07:22
I think there will be a very strong
short-term disturbance. I mean, a lot of traditional jobs
will be changed dramatically,
00:07:30
including my job as a research
scientist, as a professor. Including you as a media content creator,
right?
00:07:38
So it doesn't mean that professor's job
will be gone, research jobs will be gone,
00:07:42
and media content creators' jobs will be
gone. It's just, what we do within that job class
will be very, very different.
00:07:48
And you already see that with software
engineering, right? So these days, most advanced coders,
software engineers, they don't code.
00:07:56
They use AI to code, right? What they do is they think about the
higher-level artifacts.
00:08:02
Like, how do I design this thing? What does architecture look like? What is client requirements?
00:08:07
So they're doing a lot
more higher-level jobs than maybe the project managers
were traditionally doing.
00:08:12
So it just means that software engineers
will still play a very important role. It's just going to be very different,
just like we are.
00:08:18
Like, what I'm going to do
is going to be different 10 years from now or five years
from now than I would do now.
00:08:22
And that could be exciting for some people,
and that could be scary for some people.
00:08:26
So it depends on how you look at it. So since you mentioned it, how is your job
as a professor going to change?
00:08:32
Do you see it? Oh, I think it's going to change
dramatically for several ways.
00:08:38
Number one, as a research scientist,
like, how we research is going to change. Some of the stuff I do, like, you know,
that would take months to do,
00:08:46
Now I can use Claude Code, and it'll be
done within a couple hours, in a week,
00:08:50
in a day, right? So it brings a lot of thinking about,
"Well, maybe, you know, this idea
00:08:57
generation and verification is going to be
fundamentally changed." So how do we think about hypotheses?
00:09:04
Because now we can test hypotheses really
quickly, potentially. So we've got to think
about, like, this loop
00:09:12
between creating a problem
and solving a problem. - Right.
- And that's exciting.
00:09:17
And as a teacher, as a, you know,
professor to undergrads and BA students, I think fundamentally how we teach is
going to be changed quite a bit.
00:09:27
Like, you know, why would I come to your
class when I can have a ChatGPT to, you know, ask all the questions,
right?
00:09:33
So I think that has to be thought about.
How do we manage the classroom experience
00:09:37
so we can effectively use AI as opposed to
just, you know, being replaced by AI.
00:09:41
Lynn, always great to talk with you and
get your insight. Thanks again for a few moments today.
00:09:46
Thank you so much. Thank you. Lynn Wu, who's
Associate Professor of Operations,
00:09:50
Information and Decisions
here at the Wharton School.