Subscribe
Share
Share
Embed
Hosted by Ivey Executive Education, Learning in Action explores current topics in leadership and organizations. In this podcasting series, we invite our world-class faculty and a variety of industry experts to deliver insights from the latest research in leadership, examine areas of disruption and growth, and discuss how leaders can shape their organizations for success.
To learn more about Ivey Executive Education and the services we offer, visit us at https://www.ivey.uwo.ca/executive-education/.
CHRISTIAN DIPPEL:
Your job as a leader,
whether it's at
the C-suite level
or leading a strategic business
unit or whatever the case,
I think it's going to become
more about asking good questions
and less so about being the
person to provide the answers.
ANNOUNCER: Welcome to
Learning In Action,
where we explore fresh ideas
shaping leadership today.
This episode is a recording of
our live session in conversation
with Dr. Christian Dippel,
Associate Professor of Business,
Economics and Public Policy
at Ivey Business School.
If you're asking how to move
beyond the hype around AI
and turn it into real
strategic impact,
you're in the right place.
Christian walks us through
spotting high potential AI
opportunities, aligning them
with your broader strategy
and bringing
organizational clarity
to a fast-moving
technology wave.
Whether you joined us live or
are tuning in for the first
time, this episode delivers
"can't miss" insights.
Let's get into it.
HOST: Christian, welcome.
Thank you for joining us today.
CHRISTIAN DIPPEL: Thanks
so much for having me.
HOST: Why don't we start
out on a more personal note,
and if you could share a little
bit about your personal journey,
which spans the rigor of
research and academics
all the way through to practical
applications of organizations
that you're directly
working with?
CHRISTIAN DIPPEL: I've been a
professor for about 14 years
now, graduated in 2011
and spent the first 10
years of my academic life
at UCLA in California.
Really, at the beginning
of my academic career,
I wore three hats.
I had an investment
banking background,
so I had a bit of
that finance hat.
My formal PhD training
was in global macro,
really international
macroeconomics,
and economic history.
And so those were my
three hats that I wore.
And then over time
through my career,
two things influenced me
and pushed me, so to speak,
a little bit towards the path of
what we're talking about today.
The first is being at UCLA, it's
a very entrepreneurship-heavy
business school.
So among North American
business schools,
it's got the highest
ratio of MBAs
starting their own business
after MBA graduation.
So you get a lot of this
Silicon Valley or Silicon Valley
adjacent Southern
California tech
entrepreneurship,
venture initiation stuff
that you get exposed to.
The other thing that happened
is I became more and more a data
scientist, and I
got increasingly
frustrated with the fact that
in academia, what often happens
is people engage with these
incredibly interesting data
science projects, but everything
is on the academic funding
cycle.
So you get a big grant
to build a cool thing.
You build the
thing, then you try
to write three or
four papers on it,
and then you move on
to the next topic.
So everything is optimized
for the paper writing, which
is fine, but it's
really frustrating
that you get these
cool data projects that
then lead these zombie lives
on the internet, where you
go on a website and you
know, this thing was built.
It was so cool.
And then it says,
last updated in 2019
because the grant
ran out, the people
got tenure or lost
interest or whatever.
And so I was of
this mindset that I
wanted to build data
science projects that
have the ability to
self-sustain, which then
naturally meant they had to
have a commercial element
to themselves to break out
of this research cadence,
if you will.
And then I quickly find
out that if you build it,
they won't just come.
So in other words,
you can't just
build the thing that
you're in love with
and assume that it can
stand on its own two feet.
And I had to quickly
learn a lot of strategy.
And then, really, it sort of
became this natural confluence
where I was building
projects in data science.
I had to learn the strategy.
And then right around the
time when that happened,
AI came around.
And so for me, strategy,
almost overnight,
very quickly just
became AI strategy.
So in other words, for
me, for what I was doing,
there was no strategy
that wasn't AI strategy.
HOST: Thank you for
the introduction.
I'm smiling because it's two
comments here at the-- build it,
and they will come doesn't
necessarily really happen.
It's how do you go beyond
that, and how do you engage?
And given the topic at hand,
even when you reference 2019,
when it comes to AI, that
feels like it's centuries
ago in terms of how fast
everything is moving.
So let's go back to the
topic or even the headline.
We talk about, creating
a competitive advantage,
and how can I play a role?
I actually talked about
moving away from buzzwords.
Competitive advantage almost
feels like a buzzword,
to some extent.
It's obviously
something that we strive
for when you're
leading organizations
to create that edge.
So how can we think
about it maybe
in a more practical
or pragmatic term?
CHRISTIAN DIPPEL: I got into
strategy in a roundabout way,
you would say.
I wasn't trained
as a strategist.
I was trained as an
economist in global macro,
and then the
strategy came later.
And when I was at UCLA,
I had this very renowned
senior strategy colleague who
retired midway through my stint
at UCLA, Richard Rumelt. And
he used to tell the story of--
he would have another senior
colleague come into his strategy
class to maybe get into
teaching strategy himself.
And after sitting there for 10
classes, he walked up to Richard
and said, I think I can
summarize your entire strategy
class in one question,
what's going on here?
The point that he was making
is, at the end of the day,
strategy is--
it's a fairly unstructured
thing in the sense
that it's so context-specific.
And this was an interesting
experience for me
as a macroeconomist who is
really a data scientist.
You're always in the
business of generalities,
so you're always
trying to figure out,
what does the average do?
What are like the
patterns in the data?
But when you start
to think about,
what is going to make my one
thing successful in this sea
of other things around it,
it's like your whole way
of thinking totally
pivots into-- everything
becomes about context.
And the data is a
background thing.
You need to know
some business trends.
You need to have a sense
for consumer behavior,
competitive landscape behavior,
all these things that can
be expressed quantitatively.
But at the end of
the day, it becomes
this highly context-specific,
very qualitative endeavor
and really thinking deeply
about, what is going on here?
The answer to the question,
what is going on here,
is going to be different
for every kind of business.
There will be businesses
where AI really hits you
on the process innovation.
There'll be businesses
where it really hits you
on the product innovation.
And I think it's really about--
the way that I think
about it is really
to not be too heavy on
the playbooks, like, let's
apply this framework
or that framework.
Frameworks have their use, for
sure, but nothing, I think,
replaces deeply thinking
about your situation.
HOST: Great way to frame it.
I find sometimes when we
use these bigger terms--
and even I had a
recent conversation
with another faculty where
we dug in specifically
on strategy and
scenario planning
and, how do you stay ahead in
today's rapidly changing world?
It's sort of broke
it down to say--
but it doesn't have to be
this big grandiose piece where
you almost run the risk of
making it bigger than it is
and creating that inertia.
It's like, OK, so
what's going on today?
What's going on tomorrow?
Continue to ask myself
these questions.
So if we go to
your work, we look
at drawing on political economy,
institutional economics.
How do these lenses
illustrate the strategic role
that AI can play?
And maybe I'll put
in a second part
to this question around
hidden assumptions
that might be at play as well.
What are your thoughts here?
CHRISTIAN DIPPEL:
In my ventures,
let's say-- let's start there.
Those ventures, as well as
my research, really are in--
political economy people refer
to that as government relations.
It's really similar things.
And I saw that transition of--
2011 was when I got my PhD.
2014, '15 is when it
was cutting-edge to be
using keyword extraction
to characterize
certain bodies of text.
By 2016, you can show off
with your machine learning
algorithms.
By 2017, you do
sentiment analysis.
And so the bar kept getting
raised but in a somewhat,
I would say, incremental manner.
But it's really 2023
when it hits you.
Suddenly, there's
this level shift.
There's a million
things that you
can do that you
couldn't previously do.
And from a business perspective,
anyone with a lot of exposure
to data or a lot
of in-house data
or needing to grapple with a lot
of data-- there's text as data
and there's quantitative data.
I would say the
manifestation of AI
that we're talking about mostly
today, which is not agentic,
but let's say just
like LLMs, that's
mostly about text as data.
And text as data is really
an application where,
if your business interfaces with
text as data, which mine did--
with text as data,
you're constantly
in this world of not being able
to see the forest for the trees
because there's
just so much text.
So if you're interfacing
with government,
there's hundreds
of bills every year
that affect hundreds of X,
which in turn affect thousands
of regulations.
There's thousands of pieces
of formal commenting, lobby
meeting, committee
meeting transcripts.
There's just such
a big ocean of text
that no traditional method can
really get a good handle on,
and that's really where AI it
lowers the bar so much in terms
of even a very
small business being
able to do so much
with a lot of data,
where you went
from being able to,
do more or less, nothing
at all five years ago
to being able to do
enterprise-level applications
with a tiny team if you are
in this kind of very data
heavy environment,
which, in my case,
in the political economy,
institutional economics type
environment, we work.
HOST: I have
conversations frequently
with leaders who are grappling
with almost too much.
There's so much information.
There's so much data.
I don't even know
where to start.
So I'm keen.
I want to take advantage
of all the data.
I want to be able to
analyze and look at it
and get insights
that maybe I wasn't
able to get before or get them
in a much more expedited manner.
Do you have an example of an
organization or maybe someone
that you work with
that's doing this well?
And how do they start,
or what are they doing?
Are there any
secrets but are not
really secret that we
could share to just help
those that are listening in who
want to move forward, know where
to start?
CHRISTIAN DIPPEL:
There's general, almost
irony you could say, where--
I always think, in strategy,
there's two buckets of strategy,
or at least that's a frame
you can impose on it.
You can say there's
quantitative decision
making, which is operating
in data-rich environments.
So a typical example of that
would be pricing decisions.
For the most part,
if you're operating
in a market that's
well established,
you can make very
data-informed decisions
about pricing decisions.
You can call that strategy,
but it's very quantitative,
very data-driven strategy.
And then there's very
imagination-driven strategy,
which is the strategy you
apply to settings for markets
that don't exist yet.
Who knows what the
market for space travel
is going to look
like in 10 years?
We can't know this.
We can only imagine it.
And I think, one of the
tensions that people
seem to have when
it comes to AI is--
AI is all about the data.
It's all about ingesting
and transforming the data.
Yet when we think
about, what are
going to be the repercussions
of AI for my business
or in my industry-- it's really
a very qualitative exercise
of imagining the future
that is not-- inherently,
because it is the future,
can't be very data-informed.
And because you can't
really interpolate
from what's been going on in
the last two or three years,
there's just not enough data.
It's not established enough.
It's too exponential.
The process is too to
accelerating to really use
data to extrapolate into what's
that going to look like in two
or three years.
So really, you just
need to think about it.
HOST: Yeah, for sure.
You mentioned earlier around
frameworks, which sometimes
can be a challenge
if we over rely,
also can be an enabler as a
way to get things structured.
So let's play out
strategy and maybe
sort of bold or future thinking.
Frameworks that either you've
used or approaches that you've
seen that help leaders
process their thinking
or start to get things
organized so that they
can make better decisions,
they can lean into the data
to create insights--
CHRISTIAN DIPPEL: So strategy
is heavy on frameworks
and I think for good reason.
I think frameworks are
always a little bit
of a double edged
sword in the sense
that a framework
offers you a playbook.
And I think at the
end of the day,
you need to think deeply
about your situation.
And I think playbooks
or frameworks
can help you get there
quicker, or they can help you--
It's almost like,
you know, you're
trying on a Phillips screwdriver
and a flathead screwdriver,
and you see which
one fits the screw.
So you have tools
at your disposal
that helped you
solve the problem.
Where I caution always
when I work with clients,
when I work with
students in exec ed,
is it also satisfies
a little bit our bias
towards not having to think
deeply and just having
a playbook to play.
If someone just hands me my five
forces canvas or my business
model canvas and I
can fill in the thing,
I can feel like I've
achieved something.
But I think ultimately, a person
that doesn't have those tools
but is really willing to sit
there still and think deeply
about their situation,
they will come up
with better solutions
than the person that
is just superficially filling
in their business model canvas.
And that's not a criticism
on the business model canvas.
It's human nature that I think
it can be a little bit offering
us sort of this like
fake ticking boxes sense
of getting a handle
on a situation.
So I love frameworks.
They're very useful.
But I always caution students or
in consulting work with clients.
They're a means to an end.
They're not the end in itself.
HOST: I like your analogy
around the screwdrivers.
You know, we sometimes
confuse motion with progress.
So someone can actually be
moving but not necessarily
advancing.
But it satisfies this.
Oh, I'm busy.
I'm doing things, right?
I'm making stuff happen.
How would you help someone
sort of shift between the two?
So are there any
cues that I can say,
like maybe I'm not doing
enough of the hard work,
and I'm sort of
relying too much on.
OK, I filled out the sheet.
I've done work, pat
myself on the back.
What are some things
I should be holding
myself accountable to maybe
or paying attention to?
CHRISTIAN DIPPEL: If you look
back in history, and that's
always like when you
look back at the past,
things appear very obvious, but
they're not obvious at the time
or for us looking forward for
a new technological disruption.
But with past
technological disruptions,
you often have these first order
effects that are quite obvious.
The radio comes around, and it
totally disrupts the newspaper
industry.
TV comes around, and it
completely disrupts the radio
station industry.
The railway comes
around, and it kills off
the Pony Express in
the American West,
and it completely
destroys anyone
who is invested in canal
building, which 10 years prior,
was incredibly profitable.
Because these are sort of first
order obvious replacements.
And many in what some
management people
like to call fast history.
So recent technological
history, whether that's
in microchips or
PCs or whatever.
You get these examples like
Netflix killing Blockbuster.
And they clearly,
Blockbuster just
didn't think about
the first order effect
of home delivery of DVDs.
And then subsequent
to that, the long game
was always going
to be streaming.
They didn't think about it.
They got killed by it.
Kodak getting
killed by digicams.
But then oftentimes,
when you think
about the really fundamental
technological changes that
really changed the way we
live, the second order effects
are actually much
more interesting
or, well, depending--
if you're Kodak,
they're not more interesting.
Depending on what
business you run,
you might be more affected
by the first order effect.
But what I'm trying to get
at is you might be operating
in a business where you say, AI
doesn't affect me all that much.
But when you think through
the second order effects,
think about something like
the tractor coming around
in the 1920s completely
changed agriculture,
completely changed employment
opportunities in agriculture.
Those are like the
first order effects
that people could
see right away.
But then within 10
years, what happened
is you got massive
rural urban migration,
and you got complete changing
of the urban landscape,
the rise of superstar
cities, urban manufacturing,
the rise of the service sector,
and just this like infinite
plethora of new
opportunities coming around.
And that's the part
where I sometimes
think, because I
was fortunate enough
to have some training as
an economic historian,
you have a bit of this
longer run lens of,
OK, let's try to really think
through what this thing is going
to do in five
years, in 10 years,
and not just get hung
up on the difference of
does my competitor
have a chatbot today,
and I don't have a
chatbot, and what
that's going to do to my
perception of my brand
in the marketplace today?
HOST: So let's build on
that last point, right?
And I think most leaders
endeavor to learn from the past.
Right.
We can't change it,
but we can certainly
glean insights from it.
So how do you challenge
and get someone
to think five and 10 years
when we're talking about AI?
And I'm thinking,
jeez, something's
going to change in
five or 10 days.
And that idea of,
oh, I don't have
a chatbot that is on my site.
And if someone goes
to my site, I'm
going to look like
I'm behind the times.
So how do you
toggle between what
do I need to do right
now while anticipating?
Because if I don't think
five years plus out,
I might be in for a much
bigger surprise than not having
a chatbot today.
CHRISTIAN DIPPEL:
There's a lot of myopia
in how humans think in general.
I think the value
of trying to bring
a little bit of economic
history into whether that's
a one on one consultation or
whether that's a classroom
setting.
Bringing a little bit of
that historical perspective
can allow you to do a
little bit of back casting.
You can say, OK, what is
it that people didn't see
in the '60s or in the '70s about
what the arrival of all sorts
of home appliances would
do to female labor force
participation, and then female
educational attainments,
and then changing
gender roles 10, 20,
30 years out from those
technological innovations?
What are the kinds of
biases that are making
it hard to see out like that?
Really trying to, I think,
lean into just sitting
back and imagining what is my
industry going to look like,
either in free form or
doing that through something
like Porter's Five
Forces framework.
I think that's really important.
HOST: So how do you
feel about using
AI to cross-check your own
human-based deep thinking,
right?
So you've sort of challenged
to say, carve space.
Push yourself.
Exercise that what's
going to happen
to my industry in my space
five years and 10 years?
How can AI help me do that?
So I'm still going
to do the hard work,
but maybe it can fact check me
or cross-reference me or keep me
on track.
CHRISTIAN DIPPEL: AI means
many different things
to different people, different
strategic business units.
My own perspective
and a little bit
that's shaped our
conversations so far
is when you're operating
in a kind of business that
is more data-rich, there are
a lot of operational things
that you can do with AI.
One thing that I'm finding in
just some work with clients
is there's a lot of small
and medium-sized businesses
that have a ton of data.
That could be
manufacturing data.
It could be order
processing data, client
facing data, all types of data.
But up until now, it's
never been worth hiring,
let's say, two data scientists.
Like if you're not like
a data-driven enterprise
per se, you just sort of
accidentally, you might say,
or just as a side product
of your business operations,
I think a lot of
enterprises are sort
of moving into a world
where you've accumulated
an incredible treasure
trove of data,
have never really
done anything with it,
and now, it's a matter
of can I build a back end
maybe in a very
sandboxed way that
doesn't take a lot of resources,
where I can really come up
with new business offerings
or just really process
optimization,
maybe knowing which
clients are due a maintenance,
which clients are due,
a new product,
having an automated
email pipeline
that automates part
of the outreach
or standardizes some
of the procedures
we haven't standardized
in the past.
So that's the part where
even as a small business,
you can use AI to
automate a lot of things.
The thing that I think
you're alluding to more
is, is almost like a slightly
more individualistic enterprise,
which is you might just
sit there at your desk,
and you have to make a decision.
And you can go speak to your
colleagues at the water cooler,
and that's useful.
But maybe you're
worried that everyone
is drinking the same
Kool-Aid to an extent when
it comes to certain
strategic business decisions.
Now you have the
ability, and that's
a very individualistic
exercise that
is almost something that I
think in a business culture can
be encouraged.
But early on, it's not
really actually that clear
how to even standardize it.
Encouraging could be
something as simple
as the company will pay for
your $20 a month subscription
to one of the models
to have a more premium
access to the capabilities, like
deeper thinking, deeper research
capabilities.
That's not a lot of
money you need to spend
to use those kind of tools.
But what they allow you to do is
really check your own thinking,
scrutinize your own thinking.
And of course, any
individual can use that.
You can use that to
prepare for job interviews.
You can use that to prepare
for this particular interview.
I didn't do it, but
I could have tried
to characterize Brian
Benjamin to a chatbot
and ask what kind of questions,
is he going to ask me?
And so there's a lot of
capability to do that.
And one area where I see that
a lot is in the classroom now.
Just in the last, let's say,
12 months, the ability to just
use really any LLM, I was
hesitant to just say ChatGPT.
ChatGPT is becoming a little
bit like Xerox for copying.
And getting a really
good strategic partner.
So when I teach geopolitics, I
might throw a question out there
like, OK, imagine
there's a scenario where
we are an electric vehicle
producer in Ontario,
and there's a civil war in the
Democratic Republic of Congo
where all the cobalt comes from.
Chances are zero people
in the audience will
have deep knowledge on this.
But you can do a
scenario analysis.
You can describe
the thing to ChatGPT
and do a scenario analysis
within five minutes.
And within five minutes,
we can have a group
strategic discussion.
So we get elevated in
our industry knowledge
and in our sort of high level
strategic thinking and an issue.
Within minutes, you can
get elevated so much.
And then of course, that
can't be all and say all.
There needs to be additional
guardrails around the process,
and there's all sorts of things
you need to guard against and be
worried about.
But as a quick way of just
getting deep into a topic,
scrutinizing your own
thinking, it's just incredible.
Predictive nature can
be an interesting tool
to grab from the
past and context.
And so forth.
How can AI be used to help
me imagine 10 years out
if I am trying to see
where my industry is going
or what's going to happen?
Is it a tool I can use
in that kind of capacity?
CHRISTIAN DIPPEL: You often
hear this one liner when
it comes to AI, and I think
there's a lot of truth in it,
that it really raises the
value of good questions.
In other words, your
job as a leader,
whether it's at
the c-suite level
or leading a strategic business
unit or whatever the case,
I think it's going to become
more about asking good questions
and less so about being the
person to provide the answers.
And so there's a lot of
debates in the AI space.
Like with my companies,
I interface more
with the sort of
technical elements
of prompt engineering
and automating
prompt engineering through the
various LLMs, APIs, et cetera.
So asking good
questions is becoming
more and more important.
Some of those questions
will be automated questions.
So if you just have a data
feed that operationally matters
for your company, you want
to have an automated pipeline
in the back end that
queries the data feed
and generates the
answer from the data,
and it's all about
asking good questions.
That's sort of like
the engineering side
of asking good questions.
But I think at the
more qualitative level,
if you're just a person
making a decision,
you're sitting at
your desk, and you
can use AI as a copilot in a
non-technical sense, so not
as an explicit copilot but just
as someone to run ideas by.
So much of the way
we ask questions
is informed by our context.
And I'll give you one example.
When I teach an executive
education class in Canada,
and we talk about how is
the labor market going
to be impacted by AI?
One thing that you often
get is a lot of people
assume that one
area that is going
to be relatively
unaffected by AI
is old age care because
that is something
that requires the human touch.
And that might be entirely true.
But what I found really
interesting is the contrast
to if you talk to people in East
Asia, let's say in Hong Kong,
about AI, that is their
number one use case for AI,
is old age care.
Because the demographics are
so fundamentally different
in those two geographies.
We have relatively
young demographics.
Of course, we talk about
aging and things like that.
But it's not a very pressing
concern in a macroeconomic sense
when you look at our population
trees in North America.
When you look at
population trees
in Hong Kong, Japan,
Singapore, oh, my gosh,
they look very
different from ours.
And old age care is viewed as
the number one-- well, maybe not
the number one, but
one of the number one
use cases where I can actually
help in the form of literally
AI-informed robotics.
And so it's just striking
how people's predictions
about future use
cases are so socially
informed by the context.
And I think AI can help you
question your own assumptions.
In when you're asking
those questions.
HOST: You're right.
We kind of naturally
go to what we know.
And one would hope
that we can learn
from each other because as
things shift in North America
and aging does become an even
bigger sort of prominent piece,
what have we learned from
other parts of the world that
have already been there first?
I wrote down, you're
asking good questions,
and we talk about
prompt engineers.
I'm fortunate to work with
a lot of exceptional coaches
here at Ivey.
So I've had the
pleasure of being
asked really good questions.
Very specific, often open-ended,
leaving space for conversation.
And so I think that your
comment and maybe the art
of good questions is going to
be an even more prominent skill
that leaders are going to
need to hone and to build
and to refine.
So I'm going to go back a
little bit to the conversation
around sort of executives and
maybe even park executives.
I think leaders at multiple
levels in the organization
are involved in
strategic planning.
So yes.
It can be organizational
strategic planning,
but it also be departmental or
even team strategic planning.
So when we hear about AI
initiatives, most strategic
planning exercises,
we got to leverage AI
to some capacity or another.
So let's toggle between is this
sort of a process innovation?
So can AI save me time?
Can it save me money?
Can it create speed
and efficiency
with you sort of reshaping
maybe our business model
or our team model altogether?
CHRISTIAN DIPPEL: I used to
have a little bit this idea,
I think, early on in
thinking about AI,
that product
innovation is somehow
better than process innovation.
Process innovation, it
feels a little boring.
It's sort of like you're
wringing the towel dry,
like getting the
efficiencies you can get,
but product innovation is
really the exciting part.
But I've really changed
my thinking on that.
And actually, some of our
colleagues in Hong Kong
had a helping hand
in a discussion.
I remember well where
the framework that
was applied in that
conversation was more
about where is AI a
homogenizer, and where
is it a differentiator?
And the chatbot is a great
example of a product innovation
that is not going to
be a differentiator.
Chatbot technology is a very
commoditized type of technology.
I've been involved
in building chatbots.
It's a very
standardized product.
And yes, there's better
ones and less good ones.
And yes, we've seen
with Air Canada,
we've seen examples
of it horribly
backfiring because
they didn't execute
very well on their chatbot.
But at the end of the day,
it's not a differentiator.
No business will stand
out for having a chatbot.
Some businesses might be
worried that in the short run,
they might stand out
negatively for not having one.
I think those are
relatively minor concerns.
At the same time, sometimes,
the process innovation
is your real differentiator
and your real business model.
So when you think about famous
examples like IKEA, Walmart,
to a lesser extent, the network
dynamics of Netflix or Amazon,
those are really all businesses
where the business model
is on the process innovation
more than the product
innovation.
And so that is really an area
where my thinking has changed
from thinking product
innovation somehow
better than process innovation,
to really being agnostic
about the two and
thinking about where is
the true competitive advantage.
Is it plausible for us to build
a product that really sets us
apart, or is it maybe
the case that there's
something in our
processes that is
going to really be able to
leverage AI in a way that
sets us apart?
HOST: We're often looking for
the bigger splash or the bigger
bold innovation, and
maybe it's actually
something a little
closer to home
that we need to
pay attention to.
I want to dig in
because you're involved
in a lot of really cool ventures
and sort of consulting work.
A couple of live examples of
where AI is being strategically
applied.
So it's not hype.
So don't tell me
about the ChatGPT bot
that Company A implemented.
Something that be
interesting for our listeners
to hear that you've been
involved with firsthand.
CHRISTIAN DIPPEL: One
example that comes to mind
is in one of my own
businesses, which is really
a newsletter business, that
takes an enormous amount
of government text.
And we knew with that
idea that no one really
wants to have a newsletter
that is everything
that is going on in Ottawa.
No one really needs
to know everything
that is going on in Ottawa.
The value really
comes from saying,
if you're in oil
and gas, you want
to have a newsletter
that says everything
that's going on in Ottawa
that affects the oil and gas
industry.
If you're in utilities, you
want a newsletter for that.
If you're in telecoms, you
want a newsletter in that.
If you're in finance, you
want a newsletter for that.
And so we had this idea, but we
had this forest for the trees
problem, that it was,
well, government just
generates too much text, like
the editorial team that it would
require to like, parse all those
things into 20 different topics
is just enormous.
And it's a good actually example
of this process versus product
innovation, and how fuzzy that
line sometimes can be with AI
is we have a very,
very simple AI stack.
All it does is take all the
data that we're ingesting
and funnel it into
20 plus funnels.
The technology is not that hard.
It's not a particularly
sophisticated model.
It's using very off
the shelf tools.
But it allows you
to go from 0 to 1
in being able to not being
able to do it at all,
to being able to do it in
an extremely automated way.
And so now, we're in a place
where I sit there on a Sunday
night, editing 24 newsletters in
the space of an hour and a half.
Everything else is AI
generated, and I just
look through them, do a little
bit of polish, and that's it.
So your capability
as a small business
to do an enormous
amount, obviously,
depending on your
line of business,
it's harder in hardware
than it is in software.
In certain applications, you get
completely supercharged in what
you can do, and a
lot of it can be
much more about the
process innovation
than the product innovation.
HOST: Yeah, interesting.
You think about an
hour and a half.
What would it have taken
without AI support?
How long would it have been?
CHRISTIAN DIPPEL:
Well, it's almost
like it's your process
is changed in a way
where it would have been
straight up impossible
because you would have had to
see the forest for the trees.
But there were too many trees,
and you could have never
done it to begin with.
Then there is a maybe
intermediate solution
that's like six years
ago technology where
you define a set of keywords.
You parse through all the text
based on buckets of keywords
and to try to do an
initial funneling.
But oh, my gosh,
you would have had
a lot of work left to do
to then make that good.
And then you need
to still summarize
all the content, which four
or five years ago, would
have been very hard to do.
There would have been a
lot of hallucinations,
the quality of the text
would have been horrendous.
And so all of those
problems just go away.
And you're left with just
polishing, editing work.
HOST: Yeah.
So saving time on something
that you might not
have even been able to do
because of the complexity
involved.
So what's one single takeaway
that you would leave, something
that I can do as
a leader, when I'm
thinking about AI in
the context of creating
competitive advantage?
CHRISTIAN DIPPEL: I
think one big takeaway
that I think a lot of
businesses, especially
businesses that interface
with Ivey a lot,
I think have
traditionally operated
on a model where a
wait and see approach
is often the right approach when
it comes to new technologies.
And you know, in academia,
the fancy language
for that is like the real
option value of waiting.
And oftentimes, with
new technologies,
the idea is, well, some
standardized gold standard
is going to emerge.
And maybe I need to wait for
two or three years for the dust
to settle.
But then there'll
be an off the shelf
product that will do
what I need to do.
And so it's better to wait.
And especially in
the world of atoms,
like in the physical
hardware world,
that is often the case, right?
We all know Moore's law.
The cost of computing
decreases exponentially.
Let's say things like
the cost of solar panels
has also been
decreasing exponentially
over the last 10 years.
So lots of technologies.
Absolutely, it makes
sense to wait and see.
I think AI is really
different, and it's
different for two reasons.
I'm sure it's different for
more than two reasons, but two
that are sort of
prominent in my head.
One is it just
keeps accelerating.
And so if you wait
and do nothing,
the thing is going to accelerate
away from you in a way
that you might
come to regret two
or three years down the road.
And then the other
thing is the cost
of an option is just
so darn low compared
to most other technologies.
All it really takes is
start using the thing
in a potentially
very low tech way,
maybe designating two
or three critical people
in your organizations to
say, explore some options.
Free up one day a
week where you say
today is like your moonshot
research day, where you just
think about things
we can do with AI.
So the barrier to
adoption is really so low
that I think it
really makes sense
to engage today and not wait,
even for organizations that
have been well-served in
the past, with taking a wait
and see approach when it comes
to adopting new technologies.
And then the second
thing is going
to be quicker because
I already said it.
I think a lot of people
I find in conversations
grapple with this tension,
that this idea that AI
is all about data.
But thinking about
what AI will do
is all about
qualitative imagination.
And I feel like that creates
a real barrier to engaging
with it, and I think getting
comfortable with that,
with the idea that, yes, AI is
all about transforming data,
but that doesn't mean that
you can't just sit there
and think carefully
about what AI
will do to your business
or your industry.
HOST: And Thank you for sort
of making the comparison
to sometimes, it did make
sense to wait and see,
let the dust settle, let
the kinks get worked out.
And whereas other
cases, I sort of
pictured this train
going down the tracks,
and it's picking up speed,
and it's like, I'm not ever
going to catch up.
I gotta jump now and get on.
For individuals, do
you feel it's worth
paying for premium AI models.
And if so, or if not,
do you have a favorite?
CHRISTIAN DIPPEL: One of the
things that I found most amazing
is in working with engineers.
When you use something
like GitHub Copilot,
GitHub Copilot is like
integrated with five or six
different AI models, and just
seeing an engineer switching
between models
and one model will
be incredible at
changing your code.
Another model will be incredible
at giving you code from scratch
if you're just describing
what you want to do.
And so there's a
lot of nuance, and I
think the deeper down
that rabbit hole you go,
the more those nuances matter.
But then at the end of
the day, it's up to you
to decide if those
nuances matter to you.
So I routinely use
four different models
just in conversation.
I have premium
subscriptions to two.
I don't have premium
subscriptions to the other two,
and I just experiment.
And certain models I find very
little difference between them
depending on the context too.
So if you just ask,
here's the things
I have left over in my fridge.
Give me some suggestions
for dinner tonight,
it's not going to matter all
that much what you use it for.
If you want to do like
really deep thinking
analysis of an industry
profile, maybe there
will be some real
differences between a premium
model and a non-premium model.
HOST: So what I take
away is it depends.
CHRISTIAN DIPPEL: Yes.
It depends.
And maybe further
to that point, I
don't think you need
professional advice
on the decision of whether to
spend $20 a month on a premium
model.
Just spend $20 on a premium
model, and if you don't like it,
unsubscribe a month later.
It's not an
irreversible decision.
HOST: You know, someone might
like this particular tool
because they're using it for
this very specific reason,
whereas someone else might have
success with a different tool
because again, it's
a different purpose.
And given how fast
things are going,
I'm sure new tools will
be emerging all the time.
CHRISTIAN DIPPEL: To your point,
in a classroom setting, if we
have a session where we
actually say, OK, now
everyone go and prompt an AI.
People tend to use
very different ones.
Some people are on
their work laptop.
And so they often use
just Microsoft Copilot.
Other people use
GPT, some people
use Perplexity,
Claude, whatever.
You get a pretty wide range
of results in the answers.
But then at the
same time, if you
ask the same question
of the same LLM
5 minutes later, you can also
get a wide range of results.
HOST: It's like us, right?
If you ask me the same
question tomorrow,
I might get a different
answer depending
on what we were talking about.
Thanks for listening
to Learning in Action.
If Christian's insights sparked
new thinking on AI strategies,
don't stop here.
Head over to ivexeced.com
to explore upcoming programs
and deepen your fluency with AI.
And follow us on your
favorite social media platform
to catch more real time thought
leadership and exclusive podcast
drops.
We'll see you next time.
00:38:58,380 --> 00:39:02,000