Jobs at Risk from AI Right Now - Future of Work! | CXOTalk 889 - CXOTalk

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
AI is disrupting jobs faster than anyone expected.
Today we'll learn who's at risk,how to re skill at scale, and
strategies to separate winners from those left behind.
Our guest is David Martin, global lead for people and
organization at Boston Consulting Group.

(00:22):
I'm Michael Krigsman and welcometo CXO Talk episode 889.
Thank you so much, Michael. I really appreciate the
opportunity. I love listening to you and the
great interviews you've done, and so I'm very excited.
Give us a sense of your work at Boston Consulting Group.
I lead our people and organization business unit and

(00:44):
that is everything involving operating model, redesign,
talent and skills, culture change management.
So really all of the people related components that are now
so important on the AI, on the AI front.
What do you see are the core dynamics that are driving
changes in organizations as as aresult of AI?

(01:07):
When ChatGPT came out in late 2022, I expected a lot more
apprehension from companies to to get their hands dirty.
And one thing we're seeing is just, you know, widespread use
of the tools. We see a lot of excitement from
both employees as well as, you know, in their own lives how

(01:29):
they're using the tools. But a lot of increased
uncertainty, both at the individual level, what's going
to happen to my role in the organization as well as at the
leadership level? How do I craft a strategy around
this dynamic market? Is my business disrupted?
So a great deal of uncertainty as well that companies are are
struggling with. Could you identify any large

(01:54):
works workforce shifts that are already taking place?
Or is this all really kind of pushing into the future?
It's both, but certainly to to the first part, there are a lot
of workforce shifts. You could look at specific
categories of jobs, so specific functions.
You see a lot of pushed in the customer service domain across

(02:16):
obviously a lot of sectors. You see a lot in software
development, you see a lot in marketing.
And so you're already seeing workforce shift there.
A lot of it is OK, how do individuals in their current
role start to use the technology?
How does their role change in these spaces?
I've spent a lot of work with software developers over the

(02:37):
past year or so. I actually was 1 before I joined
BCG and you know, the uncertainty about how do I use
the tools and how do I deal withsuch a very rapidly evolving set
of tools. You see GPT 5 come out
yesterday, you know, which brings a host of new
capabilities. So you're seeing the impact
there on both the day-to-day work as well as how they're
feeling about it. There's obviously much more down

(03:00):
the line that that AI is going to impact in the workforce.
So I think there's a lot still ahead of us.
You raise a very interesting point about programmers and
software development, because I think many people were under the
impression initially that the kind of jobs that would be
automated would be, I'm going tosay, more rote or maybe lower

(03:26):
skilled jobs. But when you talk about major
impacts and software development, it makes you
realize that the impact of AI isjust everywhere.
The tentacles are everywhere. It is making folks question, you
know what their role looks like.It is impacting skilled labor

(03:46):
administrative tasks too, as youmentioned, kind of the the toil
that we talked about of folks day-to-day life.
I think very early on, historically one of the the
roles that that was supposed to be most resilient to AI was like
in the field of psychology and therapy.
And you saw data emerging in mid2023 that actually augmented

(04:09):
psychology was showing great benefit.
And so it really has been surprising the breadth of of how
it's impacting individuals role and how it's helping many roles
that maybe didn't expect to to be using AI so quickly.
You know, in terms of AI adoption, Greg Walters on
LinkedIn asks a question. He says there is a statistic

(04:32):
making the round stating that 46% of AI implementations are
failing and abandoned. And I'd be really curious just
to actually see that. But he's saying there's also
another stat that 40 billion hasbeen invested in AI by companies
but resulting in only an 8% increase in revenue.

(04:55):
So it seems like at least according to what he's saying,
that there's this tremendous investment but not yet showing
huge amounts of value from a people perspective.
Do you have any reactions or thoughts to that?
First of all, the stats don't surprise me.

(05:15):
Maybe I'm a little surprised it's only 40 billion, but the
notion of either a, some companies are struggling to
realize the value and so they'recalling it unsuccessful or be
that that the time to value is long.
So not necessarily that they're failing, but it's just, you
know, not as not as quickly as as they might want it.
It comes for at least two reasons.

(05:37):
One is companies did a really good job of decentralizing
experimentation and AI. And so you saw what we call 1000
flowers blooming and every different part of an
organization was testing it in some place.
So you saw a lot of initiatives and what our data shows is
actually the companies who are showing success and positive ROI

(05:59):
right now are actually doing fewer things.
That was surprising in our research, fewer, larger, more
focused has been faster to valueand more successful.
But I do think the other side ofit is a timeline thing I'm sure
we'll talk about. And I've, I've heard you talk
with many of your guys previously about data quality
and really just some of the the necessary enablers that an

(06:20):
organization needs to actually get the most value out of it.
And so a lot of companies are having to now play a little bit
of catch up on just getting a ship in order.
The other thing I would say around delay or maybe even
decreasing the magnitude of benefit, which does get right to
the heart of the people. Point is adoption of an employee
workforce is not always what youknow the the leaders and the

(06:42):
decision makers have expected. We saw actually the software
company, 80% of the engineers had actually expressed
excitement about using the new tools.
When we actually looked at how their initiatives with deploying
some of those tools is playing out, only 20% of the engineers
were actually using it. So it's like even if that 20%
has improved their productivity 100% and you're still at a much

(07:04):
smaller amount of benefit than the company might have expected.
And, and so adoption challenges have probably been the most
important lever that companies are now using or struggling with
and then addressing to capture value and, and I think the
problem is, and I'm sure we can dig into it later.
So sorry for the long winded answer, but one of the problems

(07:24):
with adoption is either they don't have the skills yet, they
haven't been taught properly outof prompt or they haven't even
been trained on how to use the tool.
That's one side the skills and one other component is
leadership. And you know, have they
communicated a vision on how they expect the role to change
or how employees should be usingit?
What is the what's the purpose of why they're rolling it out?

(07:47):
Are they trying to save costs? Are they trying to increase
output? So all of those types of people
related issues are many different factors diminishing
adoption and and consequently, you know, companies might be
struggling for for those reasonsas well.
This is comment question from Simone Joe Moore on LinkedIn.
And it's, this is fairly large. And actually, folks, I'm going

(08:10):
to ask you to keep your questions on the shortish side
so that the host that's me can more easily sort of read them.
But here's what she says. It's a, it's a very interesting
point. She says there's a growing
concern of employees getting tooattached personally to the
company AI or using their own AI.

(08:32):
And therefore, there's 2 problems that arise #1 they're
putting inherent bias into the system and sharing far too
personal data, and there's not enough guidance.
And #2 is when they leave the organization, they're exposing
that organization and themselvesto mental health, separation

(08:56):
anxiety, to the AI that they've been using.
It does also point to the fact that training is not simply how
to increase the output of the activities you do day-to-day,
but training is also how to manage risk of AI.
So are you introducing biases, not just about the personal
information you share, but you know, if you're in talent

(09:17):
acquisition, are you introducingbias to the process?
So one is around training on risk, I think is incredible and
probably a place that companies have under invested and then
training on, you know, how the AI works and how it's using your
data. And I think the more we see
employees understand the mechanics behind the models, the
more we see them using it in ways that probably help address

(09:40):
the first piece of it. I would also encourage folks
like, you know, if you're aware of it, which, you know,
obviously great with the question.
That's the first step is, you know, being aware of it is, you
know, there's a lot of places insome of these tools like system
instruction, custom instructionsinside of GPT, where you can
actually do a purposeful job of managing what you're sharing

(10:02):
that the GBT should kind of basea big part of its answer on.
So a lot, a lot of it there is about training.
I do anticipate, by the way, on this second piece or on the
emotional attachment that especially now and, and in your
Lenovo CIO interview, I think hedid a good job of talking about
now the prevalence of voice. And I do think I, and I see in
my personal life as I use voice mode on many of these tools, it,

(10:25):
it does start to establish a little bit of a, a sense of
friendship there. It's probably a good reason then
to also be using AI in your personal life and not
necessarily your enterprise tools.
And that might, that might soften the blow a little bit if
you were to separate. But I it's going to be a

(10:45):
challenge. There's not a perfect solve to
the to the second part of that, except I wouldn't encourage
personal use outside of outside of work on your own model.
I have to say that I also like the voice, and there have been
times it's like you kind of forget that you're talking to a
machine. It's kind, it listens to you, it

(11:07):
gives you know, well thought outcreative responses.
It comes with nice voices and I,I saw a G PD5 that's coming out
with more voices. And I mean, the scary thing is
it's, that's the worst it'll ever be.
You know, the tools are only improving.
And I'm sure part of that is howdo they become more more
empathetic? Subscribe to the CXO Talk

(11:28):
newsletter. Go to cxotalk.com.
We have genuinely extraordinary shows coming up.
So Simone Joe Moore comes back and she says she thinks that
there needs to be more proper HRgovernance over AI during
employee onboarding, role security, how they work within

(11:50):
their team, department and organization, as well as an
offboarding strategy to address some of these issues.
Absolutely true. And I think you see onboarding
now becoming exactly to this point, more cross functional.
Like it can't be an HR only thing.
But certainly how HR constructs learning and development and the
and the process around onboarding and offboarding is

(12:12):
critical to make sure that they're including it.
And, and probably including it more and more because some of
the other skills they might be training on as part of the
onboarding process now that AI can handle on its own.
Yeah, so, so completely agree with that one wholeheartedly.
I'd, I'd also say some of the tools that HR organizations and

(12:32):
3rd party software providers arecoming out with a employee
engagement platform. So a lot of companies, HR
departments are pushing on, hey,can we have a single pane of
glass for a chat bot to, to workwith employees and be more, you
know, asynchronous and self-serve.
And I think you'll see the onboarding process also have

(12:53):
that where employees are a little bit more willing to ask
difficult questions. And so I, I think that it's true
in the onboarding process, some of it is structure and some of
it is enhanced tools will actually help to deliver that
type of training as well. Just in order here on Twitter on
X, Arsalan Khan says as a disruptive tool, do you think

(13:18):
organizations are not adopting AI fast enough due to the lack
of vision by the executive leadership?
In many cases, I do think that executives are struggling with
the uncertainty and they don't have a concrete vision that if
they try to communicate, would be compelling to the workforce.

(13:39):
So I do think articulating the vision is 1.
What's fundamental to the question is, are they thinking
about AI disruptively enough? And I would say no to that as
well. A lot of companies right now are
challenging with near term issues and cost pressure and
inflation in geopolitical. And so a lot of the uses of AI

(13:59):
in the initiatives that companies are pushing on are too
incremental. It is how to use AI to further
automate tasks or specific partsof a workflow.
And they're losing an opportunity to actually rethink
the workflow end to end. And I mean, that might mean
consolidating roles. It certainly is how humans and

(14:19):
agents are going to work together to deliver a more
seamless workflow. It's not just automating
specific tasks. And so one pitfall, the
companies have certainly experienced this thinking too
incrementally and and consequently, you're not going
to articulate an incremental vision.
So yeah, absolutely. We call that reshaping, either
reshaping a function or reshaping A workflow that might

(14:41):
be cross functional. You do have to completely almost
almost clean sheet design what that should look like so that
you can have a meaningful enoughand an opportunity maximizing
enough approach to using AI. Kurt Milne on X says that if

(15:02):
only 20% of an employee group isusing AI, and that group has a
range of skills and results, then the overall benefit may not
meet expectations. No, that's right.
That's why adoption is the biggest challenge facing

(15:22):
companies right now in terms of getting the value they're
looking for out of it. And, and frankly, if you're
rolling out tools that only 20% are using it, it's also a little
bit deflating. And I'm certainly, I'm certain
for the decision makers to see low adoption like that.
And for the employees who you know in that case, I I think the
20% might be coming from my comment earlier in that case

(15:42):
where 80% of the employees were excited about it.
Kurt Milne comes back with a really interesting question.
He says are there new cognitive skills needed to use AI?
For example to run, you run AI three times and you get 3
different outputs. Then the need to spend time and

(16:04):
mental cycles reading and comparing to pick the best
answer, and that is a different skill than copy editing.
The path on that one specifically is you're going to
start to see agents that smash, you know, feedback up against
three models and try to avoid hallucinations, which is great.
It's what I do. I'm nervous about hallucinations

(16:26):
and how I use it. So I'll take the output from one
and put it in the out there and,you know, pressure test it.
So yeah, that that is that is a skill set or competency that is
changing in some of the roles. You could take customer service
too. I think a lot of companies have
a vision that a customer servicerepresentative who right now is
looking at, you know, dashboard with multiple screens and swivel

(16:47):
chairing between them, getting instructed by, you know, those
tools on what to do. That role might completely
change to actually almost a manager of agents and and that
individual then is almost like atelecommunications network
operation center where they're looking for trends that are
coming back from the agents themselves.

(17:10):
They're looking for, Hey, when and how do I, you know,
intercept some of these customerconversations?
So the the skills of being a a manager and observer of agents
and knowing when to step in or when to pressure test what
they're doing is going to be prevalent in many jobs that
currently don't require that at all right now.
And and don't usually even, you know, recruit for that type of

(17:32):
skill set. I also think like the, there's
another one on this cognitive tosome degree that being really
eager and willing to learn from the AI.
So again, working with software developers talking about vibe
coding and talking about how to introduce AI into their
workflow. Some of the most like exciting

(17:54):
interactions I saw the developers have with the tools
is like asking a question, OK, why are you suggesting I do
that? And so really like embedding
into the day-to-day like mindsetof how do I use the tools as I'm
using them to also help instructme on what they're doing and
potentially improve how I think about and do the job as well.

(18:14):
So yeah, IA lot of new cognitiveskills.
You always hear creativity will be more valued.
And I think creativity will be introduced as part of many jobs
that currently don't demand it. Critical thinking because of
hallucinations. All all of these competencies, I
think will be magnified in theirimportance.
There's so much that you just packed into that response, but

(18:34):
one of the things that kind of grappling with right now is when
you have employees that are managing groups of agents rather
than other people as employees, it raises a whole host of

(18:58):
issues, not just about job displacement, but definitely
raises that issue, but the nature of job augmentation with
AI and the relationship of jobs to AI.
Can you kind of unpack that for us a little bit?

(19:19):
You're going to see a lot of roles either working side by
side with agents in a collaborative way or augmented
way, or roles that are having tomanage agents.
It's it's a completely differentmuscle to build.
You mentioned earlier about treating AI like people.
Your Jensen Wong talk about, youknow, the CIO is the CHRO in the
future. Moderna integrated those two

(19:39):
roles. So I do think one element of
that is really being clear on the role that the agents
playing, not just how the role might change for the human, but
the role that the agents playing.
And then that human who's working side by side with them
or looking over many of those agents, their ability to to
train it and coach it and intervene to give it more, you

(20:02):
know, prescriptive detail. A lot of how the interaction
with agents will take place is very much like a manager and a
frontline employee where it is, it's collaborative, it's
coaching, it's training. I think thankfully the
performance reviews and the worklife balance matter less for
agents, which I know it's been, you know, talked about a lot or

(20:23):
or maybe in the future, you know, we get agents who are.
Who are very interested in thosethings.
But yeah, I do a lot of components of the day-to-day
life with an agent and how employees and even frontline
employees like my customer service example earlier are now
going to become managers, I think is really important.
I think you'll also see, and again, maybe back to the
cognitive point, a lot of roles and how this is different than

(20:46):
how companies were attacking digital for the past 20 years.
A lot of roles themselves are going to be expected to identify
where to use agents in the workflow and and to be trained
on how to build agents because English is code now and you
know, some of those tools are pretty easy to spin up new
agents. So many different roles are
going to be thinking about how can I introduce agents into my

(21:08):
workflow? We have two questions, one from
Greg Walters, one from Arsalan Khan that both are questioning
the role of the C-Suite. And let me just read you both of
these and these are kind of loaded questions, but but but I
think there's an interesting point here.

(21:29):
Greg Walter says AI is the end of the C-Suite.
And he also says, aren't AI and LO Ms. an example of the fall of
centralized C level command and control, shifting the way that
we work away from 19th century managed structures?
And Arslan Khan says on Twitter X, if executive leadership has

(21:56):
little vision, then what chance of success does the CIO even
have? AI is, again, like any other
software adoption, which comes down to leadership and culture.
In both cases, they're kind of questioning the role of senior
leaders. First of all, to the very last

(22:19):
point you mentioned on the role of the leaders in helping
facilitate AI adoption is like the statistics on that are
incredible. We have seen in our research
that maybe surprisingly, maybe not, but as adoption of AI
increases, fear of job loss increases, like I said, maybe

(22:39):
that's intuitive. You see the power of it.
And so you see places where I might be able to, to do more of
what you do, but the role of theleader is incredibly influential
on that. So 65% of employees, if they say
their leader is not supportive or doesn't paint the vision,
then they are fearful. Whereas for the 25% of employees

(23:03):
who say their leader is supportive, it's only 15%.
And one of one of those things is being able to articulate that
vision that we talked about. And, and it's funny, I, I know
you've had other, other individuals on the podcast who
talk about the importance of vulnerability and transparency
for leaders, which is absolutelytrue.
I think it's also been really challenging for leaders to be

(23:25):
vulnerable and admit they don't have a vision for AI.
And so you do have to be able tobe both.
So communicating the vision incredibly important.
Now the role of the C-Suite. I, I do think the first question
and the second question around what does that mean for the CIO
go hand in hand? Because I mean, first of all,
the C-Suite is almost by obligation there for fiduciary

(23:47):
responsibility, but more importantly for the success of
the organization, you are going to see more of those jobs be
less based on practitioner skillsets and more based on strategic
thinking. And so I think there's a huge
role across those different functions for what is the
strategy related to the, the business objective that I'm

(24:09):
trying to solve. I think that the C-Suite might
consolidate roles and you might see new types of roles emerging.
Do more companies, as an example, integrate sales and
marketing and have a chief revenue officer like a lot of
software companies have done potentially.
So you might see the C-Suite look different.
Now, the role of the CIO, if yougo back to that point that the

(24:29):
C-Suite is going to be much morestrategic.
I mean, this would be the point is the CEO who's dealing with a
lot of uncertainty right now about what the strategy is
looking to a CIO to help be the thought partner there.
And I mean, the problem is, again, CEOs are facing so many
different challenges right now. I talked about inflation earlier

(24:49):
in the multitude of those that they're not as able to stay on
top of the cutting edge trends. Whereas a CIOI think is, is
actually very excited and enthused by, you know, all of
the changes taking place and thenew technology is coming out.
So the role of CIO is more important now than ever, and
their ability to help craft the corporate vision and not just

(25:10):
the technology vision for the organization is more important
now than ever. That also will require CIOs to
upskill their understanding of how the business functions
because historically the CIO role was primarily inward

(25:30):
looking, taking care of systems.And now what you're describing
is CIO slash CHRO role in a way,but not just managing people,
managing groups of agents. And managing all of the risk
that comes along with agentic workflows, whether it be

(25:52):
alignment risk or cyber risk or bias risk, reputational,
obviously a multitude of new risks and a larger amount of all
of those risks that the CIO is having to deal with and having
to support the organization on. You do.
I mean, so it's funny because inHR you always talk about the HR
business partner. And I think where IT for the

(26:15):
past 40 years has consistently struggled with the dynamic of
how do I have a tighter relationship with the business,
so to speak. Those walls are being torn down.
And, and part of that is becausea, the importance of technology
is visibly so much more important be the, the
understanding of technology is increasing as well.

(26:38):
You see, you know, whether it bea president or GM of a business
unit or a functional leader, more and more of those
individuals are more familiar with technology, more informed
on the importance of it. And so you see more of a a push
pull, like more collaborative relationship with IT in the
business, which I think is incredibly healthy.
And then, yeah, the complexity of the CIO role is, is expanding

(27:02):
your point on managing agents, discovering them, managing all
the risks, supporting the organization and all of that.
I think is, I mean, again, I think you're going to see CIO
being a, a really critical role for companies who are, you know,
realizing value from their investments.
You should subscribe to the CXO Talk newsletter so we can notify

(27:23):
you of upcoming shows. So go to cxotalk.com right now,
subscribe to our newsletter so you can always join us because
we love your questions. David, your AI at Work report
describes frontline employees, agap between frontline employees

(27:43):
adopting AI tools. 51% of frontline employees, according
to your report, have adopted AI tools and leadership has an 85%
level of adoption. Does that indicate resistance
from frontline employees or what's going on there?

(28:06):
And maybe we can you can take a step back and just give us an an
overview of that research. This is a study we conducted
globally across 10,000 employeesof all levels.
It focused on the the impact that people components of an
organization have on adoption and success at AI initiatives.

(28:27):
A lot of what we focused on was leadership behavior.
We focused on talent and skills.We did get into tool quality.
We do see, you know, are the tools are the tools, you know,
good enough to use is a piece ofthat too frontline employee.
We look at adoption across all different levels of an employee
base and we did see that not only was there a significant

(28:51):
difference between adoption between leaders and frontline
employees, but also the trend stagnated, I think, which
surprised us. The adoption of of AI tools at
the frontline actually had not increased since the last time we
were in the the survey. There's multiple factors for
that. I do think part of it is what
you said, which is there is a little bit of reticence to do it

(29:13):
unless employees are told exactly what the objective is.
Many employees feel our data says many employees feel
threatened by job loss and so they don't adopt.
But there's also like you can deaverage that answer frontline
employee. In many cases, frontline
employees have less use for AI than leaders do.

(29:35):
You know, you're looking at retail store employees and field
technicians who are more and more using it.
But you know, there's a lot of on the ground labor that takes
place where AI is not going to be an important a piece of
technology for the the employee to use in their day-to-day
lives. I'd say the other, the other
thing it highlights, which, you know, keeps me up at night, is I

(29:58):
do think that leaders have observed that that potential
disruption to jobs and potentialthreat of, of, of AI to the jobs
is probably as pronounced for leaders as it is for frontline
employees. I think there's been a lot of
commentary and, and rightfully so, that that white collar labor
and that management all the way up to the top, you know, there's

(30:23):
a lot of opportunity to automatea lot of administrative tasks
and, and make that part of the workforce more efficient too.
So I think they realize, you know, you need to stay ahead of
the curve and and get familiar with the tools.
What should business and technology leaders be doing in
order to encourage broader adoption?

(30:44):
I do think how they prioritize which initiatives the push on
has probably changed historically for digital
investment. You said, you know, it's impact
versus feasibility or time and there is a new dimension that
does get into the leader piece and does influence adoption a
lot on selecting initiatives that also reduce the toil of an

(31:06):
employee's life. And again, it may be why leaders
are using it more is administrative tasks, as an
example, are something that employees would prefer to not
have to do and to hand off to AI.
So how leaders prioritize initiatives and how they include
employee centricity in the decisions they make on where to
invest, I think matters significantly.

(31:28):
I mentioned vision and I just touched on a more specific part
of a vision, which is objective setting.
So being able to really clearly communicate what the intent of
the impact of AI usage is as as an example, the tech company I'm
working with, with their software engineers, obviously
huge opportunity to improve productivity.

(31:48):
This company has no interest in reducing the size of that
workforce or the cost of that workforce.
In many cases across companies I'm working with, they're so far
behind on their digital road mapthat they are absolutely looking
for quad code and all of the different tools that help cursor
all the tools that help support software engineering as a means

(32:09):
to accelerate their road map. No interest in cost savings on
that front for this company, butbecause they haven't said that
specifically, most of the frontline employees we talked to
are worried about that and we see their adoption not be what
we think it could be. So that's another piece of
leader. The the third one that came out
in the research is leaders modeling the right behavior.

(32:31):
And I think that comes in two forms.
One is using it themselves, which we do see leaders doing.
And that we had this funny, I was in India last week and I was
meeting with Aceo and his team and they were preparing for a
board meeting and they created custom GPTS for different board
members. And I mean, you could debate
the, you know, the cost benefit trade off and all of that.

(32:52):
It was a hilarious use case thatI'm imagine a lot of the C-Suite
executives are starting to learnhow to do.
And I was like, you need to tellyour employees some of these
stories so they understand that you're embracing the technology
too, and that you're taking the time to upskill yourself on it.
And the the last piece, I was somodeling the right behaviors.
It also comes in the form of communication.

(33:14):
And are you setting, you know, apositive tone around AI and the
benefits that it can have on augmenting, augmenting the
workforce? The last one is investing in
training. If if you are a company facing
cost pressure, you're looking to, you know, reduce people
costs with AI tools, then you need to be reinvesting some of

(33:35):
that productivity savings in training and upskilling.
And if you can cut 25% of time spent, you better be applying 5%
or 10% of that back into training the employee, the
employee base. And we see it has a huge impact
on employee morale. We see it has a huge impact on

(33:56):
the effectiveness of adoption. And it's great signalling to the
workforce that that you're not just doing it myopically, but
that you have more of a plan around it.
So upskilling is another big piece that leaders can really
invest in. At the same time, yes, leaders
should be investing in reskilling, but there is a lot

(34:19):
of temptation to use AI purely for cost cutting purposes.
Let's get rid of, you know, it'sindustrial automation only.
It's done with agents as opposedto, you know, people, you know
machines. That's been true, you know,
since predictive AI to your point and you know, finance
organizations have used it to improve forecasting and reduce

(34:42):
some of the people cost they putinto forecasting.
So undeniably that's true. My point there is they need to
be reinvesting some of that savings and you can't as we
would say you know how do you cash the check on productivity
if it's in cost, you need to be reinvesting some of them.
Elizabeth Shaw has an interesting question and she
says you talked about rethinkingworkflows.

(35:07):
This is not a quick process and is fraught with business danger.
For remember, there were plenty of organizations that
experienced business emergenciesand failed projects when quote
UN quote redesigning workflows for digital transformation and
ERP. Being a former product manager
and software engineer, this one touches on the software

(35:29):
development life cycle. And they say, you know, how does
the bill become a lot? Well, you have market research
and consumer insights, customer experience work.
You got product management, UIUXsolution architecture,
enterprise architecture, DeVos front end, full stack
engineering data. Like everyone here probably
knows all the SDLC. And we've had some organizations

(35:52):
say, how do I turn that eleven person team that I don't really
know if it's a 2 pizza box team anymore.
How do I take that eleven personteam and do it with three people
using agents? That is a disruptive change
exactly to the question. That would be a disruptive
change to that workflow that is rife with risk.

(36:13):
I would say the first one is just quality risk.
Like can those three people actually sufficiently do those
jobs at the same level of quality?
And then what would the downstream impact be by having a
lower degree of quality? But you absolutely, if you're
thinking about it end to end, have to have a plan.
You need to do it so that your near term decisions, you're
making an AI investment and withyour people, you're not going to

(36:36):
try to rewind those later. You need to have the plan, but
call it a three-year end to end vision.
And then you need to be looking at what are the proper steps
along the way to get to that plan that allow for, you know,
some of the risks that you describe, whether it be skilling
risks or cultural risks by, you know, that type of disruptive
stuff. I think you'll see org

(36:57):
restructuring too as part of that I mentioned earlier kind of
sales and marketing. You know, you could think about
engineering a product, things like that.
Companies are not just going to be changing the workflow and how
people like the ways of working,but also will be rethinking
their organizational structures as part of that.
Or if they don't, I think it it increases the risk for sure.

(37:21):
Microsoft Research recently cameout with a very interesting
report where they described jobsthat are most likely to be
affected or displaced by AI and jobs least likely.
And the top jobs that they said were going to be displaced are

(37:41):
interpreters and translators, historians, passenger
attendants, and we could say parking parking lot attendants.
I'm sure we've all been to automated parking lots.
And you have to pull out your phone and try to figure out how
it works. Sales reps.
And among the the least likely to be affected are dredge

(38:04):
operators and lock operators. You know, locks like on bodies
of water. I live next to Volter's lock on
the Thames in Maidenhead, England.
It was one of the most exciting things in my day.
I was watching the locks work. I can imagine that is a
difficult one to replace. Yes, I, I would say also like to

(38:26):
to pile on the ones that I'm sure will be in increased demand
is if you look at the amount of infrastructure investment that
the US and other countries are making on their power as well as
in semiconductor manufacturing, there's a lot of both blue
collar and white collar work that's needed there that will be
I think immune to the the impactof AI.

(38:48):
So I do think there's a lot of manual task labor at the front
line that we talked about earlier that is highly immune,
as you mentioned, and I do. That's going to have an
implication, by the way, on education generally.
And I know we've talked about, you know, how the education
system works for a long time now.
But but because those jobs are going to be an increased demand,

(39:09):
we're going to need to rethink how we're building skills and
building career paths so that wecan encourage individuals to
step into those roles. There are many jobs, by the way,
that will be significantly impacted by AI, but it will not
lead to job loss necessarily, but actually just an incredible
explosion of, you know, work in that space.

(39:31):
You could take software engineering.
I'm sure there's going to be a lot.
I wondered if lawyers who peoplesay are potentially, you know,
there's a lot of automation there.
Will there be 100 times the legal case like you do?
You do wonder how much people use the productivity
improvements from AI to increaseoutput and activity versus just

(39:52):
replace human labor. On this topic of training and
skilling, a recent Fortune article warned that Gen.
Z should learn AI, but their jobprospects may still be difficult
because of structural changes caused by AI.

(40:15):
So what advice do you have for folks in Gen.
Z who are going to be facing this wall of automation?
You're seeing it some right now.Well, I, I believe I read that
unemployment rate for recent grads has ticked up a little
bit. And it's interesting.
We don't talk about it, but I have a 18 year old daughter
who's about to go off to college.

(40:35):
So there's a lot of conversationabout how AI is going to impact
her, you know, post college career and, and what she should
study and all those things. I mean, 11 great thing about
Gen. Z is I'm positive that they will
be highly adaptable and resilient.
They're obviously the savviest users of digital technology and
it's already embedded into most of the apps that they're using.
So at least they'll have the familiarity with the tools.

(40:56):
Now what are they going to do from a job perspective?
You're going to see an incredible explosion of
innovation and new companies andnew industries that emerge.
I just like we did with.com and I think that Gen.
Z will be on the forefront just like, you know, a lot of them
created the YouTube creator industry out of nothing.
I think you'll see a lot of really neat new industries

(41:18):
emerge from the creativity that comes from that age group.
I do think probably their careerlooks very different than what
my career looked like. But I think that's true between,
you know, myself and my father as well.
So that's, that's just the nature of time and how we
evolved. But but it is a concern and and
like I mentioned, I do think it's a concern too on where they
spend their time and how they invest in skills leading up to

(41:41):
entering the workforce. Absolutely.
You mentioned earlier this time dividend that arises from AI
automation. It raises the question to what
extent are these benefits real? So as companies adopt AI, will

(42:02):
they really migrate employees tohigher value work or as we were
talking earlier, will, will these employees simply be made
redundant and fired as part of garden variety cost cutting
measures? So how do you think this all
plays out? It's dangerous to only focus on
the cost measures. And, and, and here's why,

(42:22):
because I think a lot of companies who are probably
asking those questions are also potentially ripe for disruption
or their industry, you know, being disrupted.
And so I think you will see partof the time dividends start to
go toward innovation. If you think about the old
Google model, and it might stillcurrently be, but I, I know it
from a long time ago of, you know, spending 20% of your time

(42:45):
thinking about other, you know, blue sky initiatives are working
on it. You will see companies more
strategically allocate time fromthis productivity savings and
time dividend toward innovation.And, and it's because I think a
lot of companies are going to start realizing the disruptive
threat of, of, of disruption. Obviously, I think the other

(43:06):
piece on on time dividend, and I'll use a specific example
because you asked like, will they do higher value activities
again, going back to software development, It's going to be
true in marketing as well. And we see that a little bit
talking to a company about usingAI to improve the productivity
of the daily coding. And that individual, that team

(43:27):
actually was dealing with legacyIT that has a bunch of
interdependencies. It's all the technical debt that
in all of our enterprise listeners here deal with
everyday and they have not been able to spend the time and the
thinking power on refactoring code and how do we design this
is a Java 8 to Java 11 upgrade or something to that nature.

(43:49):
And so it's like the strategy behind improving the quality of
the existing assets that the company has, I think will be
increasingly an important role for for the time dividend that
you're describing right now. Back to the vision point, I do
think if companies aren't using it for cost savings, then many

(44:10):
companies are not being creativeenough or forward thinking
enough to not just say, OK, do more of the same.
So if you think about that example I just use, it's like
you could just say, OK, go buildmore code and you know, go
deploy more code. Instead they're saying OK, let's
go re architect. So being really thoughtful
about, you know, leaders and setting the expectations for how

(44:32):
you're using the time, I think it's really important we see
that at BCG as well. You know, we've rolled out a lot
of AI tools for our consulting staff.
They say they're saving time. You know, we wonder, OK, are
they sleeping more at night? Because for us, in a job where
our employees work really hard and work life balance is
increasingly important, it's like, great.
Yeah. Our goal with the productivity

(44:54):
is, you know, they can get more time to have a more sustainable
work life, but you need a plan for what you're going to do with
it. To your point, I don't think
that everyone is just going to say do more of the same thing.
That would be myopic. On the subject of consulting, I
recently saw a demo of agentic AI tools that are designed

(45:16):
specifically to replace junior consulting consultants doing
research and gathering information.
How real is this AI threat in consulting?
It's an incredibly helpful tool for our consultants to use is,
is one thing. So and, and what I mean by that

(45:39):
is yes, it does do a lot of the work that many of our
consultants do. So there's a threat side of
that. And then in the near term,
there's a, it takes a lot of toil out of their job and allows
them to be more strategic. I think it opens up
opportunities for new business models for us.
You know, are there opportunities as an example for
us to provide more of a softwareas a service and a consultant as

(46:01):
a service? Because we might think we can
train that type of model better than anyone else can.
That's one example. But we are tracking and, and by
the way, I think all of us partners, before we go into any
of our client meetings, are doing deep research on the, on
the client. We are pressure testing our own
analysis and saying, if I'm a client, what would I be seeing
from AI? And are we exceeding that bar

(46:23):
and justifying, you know, the fees that we charge?
So I think that it's been very important for consulting to
appreciate, you know, the, the side by side threat.
We actually have been measuring what components of the value we
provide our clients are replacedby AI, to your point, and it
stretches across many different dimensions, industry expertise

(46:43):
and all that kind of stuff. And one place that AI has not
come close to cracking the nut is our ability to understand our
clients and their needs at a company specific level very
deeply and to tailor our answers.
Right now, if you run deep research or if you're using any
off the shelf tools to try to solve the strategic problems

(47:04):
that we saw on a day-to-day basis, it's going to be more
generic and more for the industry rather than more
specific to the company. And I think that's harder to
replace. You'll of course see enterprise
models being trained over time, I'm sure.
But right now we feel like our ability to really think with the
lens of our specific client in mind and tailor solutions for

(47:26):
them is is improved because of this.
Like back to the point, I think it frees up our time to think
more about those strategic tasks.
This question about OR issue about LLMS not providing deep
company specific information. I wonder if that's at least

(47:47):
partially a data problem becauseI have seen a number of
companies recently that go out to essentially aggregate
interrogate databases across a company.
And once you start doing that, you you build up the data that

(48:08):
you need that a sufficiently fast and powerful LLM could then
use to report back whatever depth you might want on all
aspects of the company, sales, marketing, everything,
customers, you name it. Yes, and using a generative AI
as a potential mechanism to leapfrog some of the like this

(48:31):
is I think part of I think this is partly in there like using
Jenny I to actually leapfrog andmaybe mitigate some of the lack
of connectedness with company data is also an opportunity.
I think you're spot on. And and that is fed, you know,
historically into things like retrieval augmented generation
where you're using off the shelfLLM you're you're injecting some

(48:52):
first party data in there to getsomething that's more company
specific. I think that's absolutely right.
That's part of the near term future.
I think The thing is our clientsare not necessarily saying that
that is the highest priority place for them to be spending
their time. And in a world where as we know,
talent is so slim and IT talent specifically and tech talent

(49:12):
specifically and core business needs for a company are so
pronounced and there's so much opportunity to improve there.
I have not seen a lot of our clients dedicate a lot of time
and resources to what you just described because they they have
incredibly important priorities they're pushing on in parallel.
So. Arsalan Khan says I'll just read

(49:35):
his question verbatim. What about power of 1 AI in one
department versus AI in another department?
Whichever department has more power, people, revenue, will
they get their way? Who is the AI referee?
Right. Well, and you've seen that in
like data reporting and analytics now for the past 10

(49:57):
years where every organization has their own analytics team
that's creating their own dashboards.
I think there's some truth to that.
I think it, it points toward theimportance of IT and the
enterprise IT strategy and platform strategy.
I think one danger that companies are facing is the
fragmentation of platform decisions, you know, across

(50:18):
different functions. And so resource allocation, to
your point that comes from central IT and the distribution
of tech talent toward different functions is increased in
importance. There's something about human
nature there. And the question that's like, I
think of course different parts of the organization are probably
going to have more power and influence than others.

(50:41):
AI is not solving for, you know,basic human nature at this
point, for better or for worse. So that will probably be the
case. I think it's really important to
be at ACEO and CIO level though,and how the strategy is crafted
there. Because what you don't want to
do is have a lot of function specific platforms that again,
like only really harden how the organization currently works.

(51:03):
And the more you can make the platforms cross functional, the
more you'll be able to realize amore seamless customer
experience. So ideally you're not having
those siloed platforms. Your AI at Work report talks
about global differences in AI adoption.
I think this is very, very important for people to
understand. The report states that

(51:24):
respondents from the global S have significantly higher
adoption rates than in the West.For example, India and folks
listen to this has India has 92%adoption compared to 64% in the
US. Can you just tell us about the
implications of this? It seems very profound to me.

(51:45):
The implications are right and Ijust came back from Delhi in
Mumbai and it's incredible what they're doing there.
Some of it is job specific. You know, we talked earlier
about just the nature of some roles using it more than others.
India obviously has an incredibly large and strong
population of software developers and those types of
roles that are more suited to use it.
Demographic difference it does. It points out some of the

(52:07):
differences in demographics too and where some of just countries
will differ because they're either a younger workforce or a
more aging workforce. India on the younger side
relative to some of the developed markets it.
But your last point I think is the most important, which are it
does have implications on what you think the future looks like
5 years from now. Now I think India like amazing

(52:31):
users. You'll probably see a lot of
innovation coming from there because of the usage and because
of time. Other than all we talked about
there, I do think the US while behind on adoption is obviously
making very important infrastructure strategic
decisions. And so you'll still see despite
some of the lack of adoption, I think the incredibly important

(52:52):
AI capabilities coming out of the US, but it is it is
fascinating to see there's a very broad difference across
geographies on, on usage adoption fear.
India was also one of the highest in terms of their fear
of job loss with AI, which goes back to leader point their
interest in shadow. IT was also pronounced in high

(53:12):
individuals who are not providedthe right tools are using tools
on their own, which is I think scary for CIOs and highly
prevalent and we talked about that in the report.
So a lot of interesting geospecific Nuggets in there.
David, fundamentally, are you anoptimist, A pessimist, or do you
think this world is so confusingthat who knows?

(53:35):
I'm a huge optimist. I'm not going to end on a down
point. It'll be down and up real fast.
I had a daughter, 4 kids. One of my daughter's passed away
from pediatric cancer a few years ago and I viewed it so
close to breakthrough. So I'm an optimist because of
the impact that AI is going to have on science and on public

(53:55):
health. And so I think you have to be
excited just about the innovation that's going to come
through there. I think human ingenuity and all
of the past data we have in terms of job loss and recreation
is like, we're an incredibly resilient population and
incredibly creative population. So the doomer side of it, I, I'm
very optimistic. I think that we continue to

(54:17):
thrive. We, we find the right ways to
use AI as a helpful tool to makeour lives better.
So very, very much an optimist. And with that, a huge thank you
to David Martin. He's global lead for people and
organization at Boston Consulting Group.
David, I can't thank you enough.Thank you for being here with us
today. Thank you Michael.

(54:38):
I've I've loved loved the time today and really appreciate what
you do. Your your podcasts are
fascinating. Encourage folks to sign up.
So it's great. Yes, yes, folks, before you go,
subscribe to the CXO Talk newsletter.
Go to cxotalk.com. We have genuinely extraordinary

(54:58):
shows coming up. You just need to look at the
newsletter to see what we have coming up.
I mean, really great ones, everybody.
Thank you for watching again. Thank you to David Martin, and
we'll see you again next time. Take care everyone.

All Episodes

Jobs at Risk from AI Right Now - Future of Work! | CXOTalk 889

Episode Transcript

Popular Podcasts

Stuff You Should Know

My Favorite Murder with Karen Kilgariff and Georgia Hardstark

The Joe Rogan Experience

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Jobs at Risk from AI Right Now - Future of Work! | CXOTalk 889