November 20, 2024 • 50 mins
The next 10 years will be a pivotal time for the AEC industry to harness the innovations of AI and shape its impact on everyone. Join us for a conversation with Esri about how we can leverage AI right now to address the evolving needs and challenges of our industry, driving immediate innovation for our clients, communities, and workforce.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:04):
(Micah Callough)There's a lot of
deeper R&D things
that are getting really interesting,
and we're just kind of
in that balancing game of like, where are the models?
Is GPT the thing and always going to be the thing?
I'm not so sure it will be right long term.
Scott can help me with that.
(Dave Mulholland)And I think we're hitting
probably one of the biggest,
one of the challenges
or attempting to do that
with this podcast, right, is education
(00:26):
and training of what AI is and what AI isn't.
(Scott Jay Ringle)Don't be afraid to move forward.
And it comes down to a single,
small little word (00:33):
data.
Talk to us about data.
Don't be afraid to have your data repaired
and ready to move forward.
(Dave Mulholland)Thank you for joining us today.
We're excited to come today to talk
about artificial intelligence.
Artificial intelligence
AI continues to be the one of the most influential
(00:54):
yet disrupting technologies
to impact the AEC industry in decades.
In some cases, we're seeing that is impacting us
somewhere between 25% up to even 95%,
in terms of the automation of where we're at.
ChatGPT engulfed
the industry in November 2022, and here we sit
almost two years later
(01:16):
seeing the impacts of other technologies,
and other tools that are moving forward.
The next ten years will completely be revolutionized
with the impacts of AI.
The industry needs to figure out
how to harness this technology, use it,
innovate, and automate
where we're moving as part of the AI revolution.
I am really excited to be here
(01:36):
co-hosting this with Scott Ringle, VHB's very own
AI lead.
Scott comes to us with a background
in the technology space, which complements
what we're trying to do with VHB.
He brings a wealth of experience, actually driving AI
and the practice in the business,
and we're going to hear a lot from him.
Scott, thank you very much for joining us today.
(Scott Jay Ringle)Dave, Micah, it's great to be here.
(01:56):
Thanks for having me on.
Looking forward to our conversation.
(Dave Mulholland)So today Scott and I are joined by Micah Callough,
Technical Director for AEC and Environmental
Consulting with Esri.
Micah,
we're extremely grateful for you
to take the time and join us today
and talk about this very instrumental
and challenging topic.
(Micah Callough)Yeah, sure. I'm glad to be here with you guys.
(02:17):
My name is Micah Callough,
I'm the Technical Director
for AEC and Environmental Consulting here at Esri.
My focus is on helping connect
kind of the AEC business
with,
what Esri does and kind of think about it
from a strategic perspective.
I joined Esri maybe five years ago now, seems,
(02:38):
gone fast.
Right before Covid and, I've helped to grow the
the sector
and the focus on AEC here at Esri in that time.
Prior to that, I had worked,
30 plus
years in the AEC business
between two different companies.
So I've spent most of my life
as a practitioner
on the AEC industry side in many different areas,
(03:00):
all over the place for all the things that AECs do.
Hey, Scott, thank you for joining us.
And why don't you tell us a little bit
about yourself? Sure.
(Scott Jay Ringle)I'll give you a quick summary here.
My journey began back in 1992 in IT,
came a long way since then.
But passion,
my passion
for AI formally began in 2017,
which I think is relevant to our conversation today.
(03:21):
After reading the now famous whitepaper
on transformer models
called Attention is All You Need.
Kind of the rest was history
after that, I was obsessed with the technology
and proceeded to hone
my skills at companies like DXC Technology,
one of the largest global systems
out there, and Google and Microsoft,
as you already mentioned.
I kind of specialize in foundation
(03:42):
model architecture and solving problems
for our fortune 500 clients
over at those locations where I was at before.
Now specializing in machine learning frameworks,
including neural networks,
and I helped drive advanced AI use cases here at VHB,
focusing on all things
AI and helping integrate it
into all of our workflows
and drive innovation
and deliver better outcomes across the board.
(04:03):
That's why I'm here.
(Dave Mulholland)Thanks, Scott, I appreciate that.
And,
I really wanted to take this session and really
break it down into like three different areas.
First area I wanted to talk about:
What is AI doing in the industry?
So this is a high level
what AI is actually in terms of disruptors,
some of the benefits,
some of the challenges that we're seeing.
The secondary that I wanted to get into
(04:24):
is really, Micah, this gets to you
and what Esri is doing.
Where is Esri going?
You know, what are we working on?
And then the last part I wanted to go into is,
Where do we see it going?
This is a challenge.
And I don't know that we actually fully
have our arms wrapped around it.
Three years, maybe five, ten, 15 years.
There's going to be some challenges.
Everybody in the industry is going to be facing it.
So let's start out with some basics.
(04:45):
Scott, I'm going to come to you first
and then Micah, I'll let you go next.
Where do we see...and for
the listeners that are new to AI...What
is the definition of AI?
What does it mean to you? Okay.
(Scott Jay Ringle)Let me give you my personal opinion on this first.
So to me personally,
artificial intelligence in this case
focusing on the extreme is a game changer
(05:08):
that enhances how we design,
engineer and build across the board.
Basically, it's
about bringing smart adaptive technology
into our workflows daily
and enhancing our capabilities.
So that's my personal kind of umbrella around this.
For my experience
AI simulates
human intelligence processes like learning,
reasoning and problem
solving through computer systems.
I mean, that's what I really is.
(05:30):
I've seen firsthand how I can shift
through massive amounts of data
to help us make better informed decisions,
anticipate project outcomes,
even suggest design alternatives, etc.
specific to AEC
so consider it your tireless assistant,
if you will, that handles the heavy data
lifting and allowing us to focus more on
creativity and strategic problem solving.
Kind of my philosophical view of AI.
(05:51):
Well, there you go.
(Dave Mulholland)Thanks, Scott and Micah. I did this on purpose.
I was like trying to get the general.
And then I know
Esri as we play in the geospatial world, what is Esri
and what is your definition of AI
and what does it mean to the listeners and
where we're going?
(Micah Callough)Well, when I think about it
and I cheat because I'm very much an, a, an AEC
practitioner, right. That's where I've lived my life.
(06:14):
My job has always been connecting the technology
and the business needs. Right.
Like really try to understand it.
And AI is not, is nothing new.
It's been around and components of it
have been around for quite some time.
It's just now starting to get that
the technology,
the demand and the drive is beginning
to start to match each other.
And coming over here on the products side
(06:36):
or really the drive for me
there was that
I think the AEC industry is sitting at a spot
where it can, it's
going to start to shift in a really big way.
And I could see that coming from the seat
I sat in before,
and I felt like coming over on
the product
teams would start to really help,
like, could start to help,
to steer that a little bit.
Esri has played in this space quite a bit.
(07:00):
And the thing that's most
like visible,
I think, to our traditional geospatial people,
is what we refer to as GeoAI,
which is,
quite frankly, machine learning,
image extraction, getting details out of it.
And that's
where a lot of people start their AI journey
that aren't practitioners of IT,
and technologies
that are deep down in the bowels of the technology,
(07:21):
but the business people,
and that's where they see a lot of value, right?
We do a lot in geospatial
with imagery and image stuff.
So that's where Esri really got
its start and its positioning.
But like all other technology
companies
that developed platforms like Esri,
including companies like Autodesk and Esri,
who are partnered together,
we're all really beginning, just like you all,
(07:43):
to start to think about how AI, is it a thing,
a monolithic thing,
but how it's something that starts to get woven into
what we do to accellerate or to improve upon it?
(Dave Mulholland)I agree with you hundred percent.
I think we're to the point where you're starting
to see this upward movement of where AI is.
Eventually, it'll be part of the daily practice
(08:04):
and who we are
and what we do as engineers, planners, and scientists.
So,
one of the questions
I get asked a lot, right,
and these are by people that do not know
or do not understand
AI, is
what are some of the most significant benefits
or areas that you're starting to see movement in,
using AI within the AEC space?
So, you know, Micah, I'll come to you first.
You know,
where do you see some of the biggest benefits
(08:25):
in terms of return on investment or,
or how people are using AI?
(Micah Callough)Right now,
the one that's actually truly in the business
that I see is the GeoAI stuff.
It's those models
that we're
beginning to put out
on the Living Atlas of the World,
that we create and train on different imagery.
We put publish them out into the world.
(08:45):
They're free to pick up for anybody
that's got an ArcGIS license,
and they can start to run them.
And those have been traditionally
being run
in our professional desktop tool, ArcGIS Pro,
kind of targeting that more data
engineer, data analytics
kind of persona and personality,
where we're even starting to see those tools
start to become able to access
(09:06):
within the online environment.
So your casual web
based users
can start to get information off of that imagery.
And why I like this the most is,
is that the AEC industry,
or really the geospatial side of things,
has relied on imagery to,
to be a major component of mapping,
basically making your road map,
(09:27):
making your street map, making a parcel map,
making maps of all the buildings.
These are all have always traditionally
been photogrammetry,
image-based type things and or survey.
And we always want to get imagery
on our projects, right?
But we could, in the old days
it was kind of hard and expensive to do that.
So we didn't always get the best imagery.
(09:49):
Well, now imagery and the technology behind
it is becoming more accessible
and readily available through drones
and through other platforms and commercial offerings
that all of a sudden
we've got a lot of imagery
and observations of what's
going on out in the real world
that we want to take advantage of,
and we can take advantage of it
in so many different ways because of AI.
(10:11):
And what why we coined at GeoAI
is because we're looking at
and extracting geographic location
meaning from those images.
Everything from, is there a car here?
Where's the fire hydrant?
Get me all the trees.
Stuff that we can do en masse
and for a lot less money, right?
Which makes project managers
a whole lot more happy right now.
(10:32):
We can do things better, faster, cheaper.
That's really the
the starting point
and the one that I see most practically being used.
That's not an experimental stage.
It's, it's truly in a production stage of use.
(Dave Mulholland)You know, Micah, I've been in the business 30 years.
I wish I would have had this tool 25 years ago
to help with my career.
So if we go back in time,
if we can go back in time and actually do that,
(10:54):
that would be amazing.
Scott, turning to you
now in terms of the benefits, where
where does
VHB see
some of the largest benefits or movement of AI?
(Scott Jay Ringle)Well, Micah,
his answer was extremely detailed and good
and mine
is going to be a little bit
more generic wrapped around
design services for VHB
and where we're really focusing right now
it is the biggest impact
(11:15):
that we see an AI currently
using within design services space.
So AI can generate multiple design iterations
and optimize for factors like sustainability
and cost,
and predict
how designs will perform under various conditions.
Those are major leaps forward also,
and how we can enhance productivity cost,
go to market, everything that's wrapped around,
design services,
(11:35):
AI has radically enhanced that capability.
It not only accelerates design process,
but also leads to more informative sand effective decisions.
So again, decision making across the board.
Everything has been enhanced through AI,
adopting AI driven tools and methodologies.
This design industry we were talking about
is to use greater efficiency
through various technologies.
We'll get into more later in this call.
(11:56):
But bottom line,
the impact that we see
the most benefit from
is specific
to the verticals
that we're in, is in the design process space,
if that makes sense.
(Dave Mulholland)And I don't know that you'd have any disagreement.
And Micah was actually referring
to their, their partners over at Autodesk.
And I know, you know,
when you think about the geospatial world
that Esri plays in and the design services side,
(12:16):
it does go hand in glove.
And I know that's a lot of effort
that Esri and Autodesk
and the other vendors I work with as well.
Yeah. So we talked to go ahead go.
(Scott Jay Ringle)No go ahead I was going to say ultimately leading
we're basically getting to better design buildings
and infrastructure, city automation,
AI processes and design functionalities we have.
I mean, that's what it comes down to.
Better design, better
buildings, better infrastructure based on AI.
(12:37):
(Micah Callough)So many more options, right.
Like you can do
and think about it
in so many more ways with all those crazy
overlays of sustainability and resiliency
that we have dealing with hurricanes
like we just had the other day,
if we need to consider these things in design
because it's our new normal.
(Dave Mulholland)Yeah, we're starting to see a lot of that movement.
You talk about the sustainability and the resiliency
(12:58):
and moving in towards the design services.
You know, Scott talked about design,
the embedded carbon calculations
and how we're doing it, how do you optimize it.
Scenario planning.
What are some of the,
what is you know, Micah,
what are some of your AEC clients
or some of the challenges
you have with implementing the AI?
What's keeping them up at night
and how do they actually stay ahead
(13:19):
of some of the competition driving this?
(Micah Callough)Yeah, the AI
term is a double edged sword in a lot of ways, right.
Because it's now the hot thing.
We are we still are talking about digital twins.
And that was a hot thing for a minute.
And for that it was big data.
And there's always a hot thing.
And I'm very happy to see that the C-Level in AEC
(13:40):
firms are beginning
to understand that business
transformation happens with technology, not,
outside of the technology.
Right? Like so it's a component of it.
So there's a lot going on there
that's driving that forward.
Some of the risks that I see when I look across
many of the organizations is in general,
(14:01):
it's the data? Like their data,
the most valuable thing an AEC has
is its knowledge and its data that that knowledge creates.
And right now most of that is created in files here.
And files,
there in somebody's
computer over there
and some SharePoint drive over there.
(14:22):
And
it's not, it's thought of project by project
because that's
the project managers drive, do
the project deliver the project, right?
But that data has a lot of value
that people don't even realize at this point in time.
And so if if the data is not organized or accessible
or something
that you can actually take advantage
(14:42):
of to interrogate with AI,
then you then have to start there first
to then be able to take the AI step.
So it's,
there's no like, magic
silver bullet
where we're just going to feed it
random stuff,
and it's going to give us the amazing answers.
We're going to have to do some data
engineering work to make that work the right way.
(Dave Mulholland)Yeah, yeah.
Scott, I know, I know,
(15:03):
data has been big for you and where you're going.
I think Micah took your thunder
there a little bit by saying data, because that's it.
Where do you see
some of the biggest challenges with the AI side?
(Scott Jay Ringle)The biggest challenge is on the data side.
So biggest challenges with AI? I actually have 5
things I'd like to say about that.
Number 1 as Micah already pulled it out
(15:23):
or pointed out data quality and availability.
That's a huge one.
AI models are only as good as the data
they're trained.
So inconsistent, incomplete, or biased
data can lead to inaccurate predictions and insights.
I mean, it's not just in this industry,
in this vertical,
it's in all of them
coming from the systems integration
world of working
with many different verticals and markets.
We see this is a common
denominator across
almost every aspect in building out AI models.
(15:46):
Integration with existing systems is another problem.
So when you incorporate AI to establish workflows
and I talk to workflows,
and legacy systems can be technically complex
and resource intensive.
Having the right teams that know how to do that,
these are blockers to people
that are trying to get into AI
in a more advanced level,
and there's ethical and legal considerations.
(16:06):
The other one I think is important is a skills gap.
Again, these are coming more
from the systems
integration side
of seeing the general common denominator, core issues
when it comes to challenges
with AI and integrating it into existing
or even new start up systems.
Skills gaps is one, another one is change management.
Overcoming organizational resistance
to new technologies.
This is not just this.
(16:27):
This could be again,
as Micah has said, there's always a new thing.
I remember to date myself
a little back in the day when the, the data people
and the
IT infrastructure teams
and the voice teams didn't get along and they started
merging voice over IP and infrastructure.
That was fun. Same kind of scenario.
So change management is an issue.
Those are my five concerns
for biggest challenges in AI.
And that's, some pretty common denominators.
(16:48):
Like I said.
(Dave Mulholland)Yeah.
And I think we're hitting
probably one of the biggest,
one of the
challenges are
attempting to do that with this podcast,
right, is education and training
and what AI is and what AI isn't?
Today I think AI actually is that great equalizer.
Firms, smaller firms have the ability
to leverage and partner with Micah, firms like Esri
(17:10):
and actually take in and really,
work with their data, actually
aggregate it,
put it into a right, structure format
and actually moving forward.
So how, you know Micah
Esri is one of the biggest providers
of geospatial in the country, if not globally.
How does
Esri do that training program
to get the message out of what you guys are doing?
(17:30):
And this is more of a
to give you a platform to say, here's where we are
and here's where we're going
relative to the training side of it.
And then specifically getting into your GeoAI.
And the roll out
where people are going to be using that.
(Micah Callough)Yeah, I mean,
Esri has a long history of having to
build its own people, so to speak.
(17:51):
When GIS first came on the scene
a little over 50 years ago,
it was not something
that was in the in the masses, right.
So Jack Dangermond, owner of Esri,
did a lot of work in building GIS professionals
within schools and education
and drilling back into those things
(18:12):
and seeding that out
so that people kind of understand what geospatial is,
not as in a software,
but as in it's a science, right?
The idea of mapping the world is a thing.
So there's a lot of need to try to push back in
to the education systems
and make sure that what we're churning out into
(18:33):
the industry
is, is going to be people that understand
the beauty of that is, is most of these kids
have been working with this stuff
more than any of us have. Us old people.
And they come to it
naturally in some cases,
I'm finding that you can pick up people that know
a lot about it at a very young age
and become very useful and effective in effect,
(18:55):
they're doing exactly what you just said, Dave.
They're using this new tool and this new capability
to accelerate
what they're doing in their career path.
And a lot of us are standing around
going, wow, that's amazing.
Initially, it's
going to look like voodoo magic, right?
But at the same time,
we have to take your risk question from before.
(19:16):
We still have to take it with a grain of salt. Right?
Like we need to make sure
that what we're getting is true information,
that it's actually matching what we're doing.
Esri does
also a lot of effort, as you know, at conferences,
at workshops we put out learning paths.
We do everything and anything we can to try to help
(19:37):
people understand
what they're doing
so that they're not opening up some software
and clicking the next button,
getting a result and not understanding
where that came from.
Which is why we say that
when we talk about geospatial professionals
getting into this area of AI,
we start to
see the developers
that we've always had kind of using geospatial.
(19:58):
And now we're beginning to see this analytics
and data engineering and data science community
that are being developed.
And where
we're starting to show
is that they need to be thinking
about how they work together.
And so we're trying to do a lot of change
management or evolution.
There's another name Esri likes to talk about.
I'm just used to change management.
But trying to help those teams come together
(20:20):
because it's not about like
taking a bunch of geospatial people and saying,
you're all going to do this now.
It's more about merging them together.
Case in point,
when we got into doing water
models and utility models back in the late 90s,
and the geometric network
came out in the 2000s,
and we were starting to model out
all these utilities.
(20:42):
We didn't do that alone with GIS people alone.
We did that with the engineers and the
the people that know the water utility
providers, right?
The people that know that business.
And together they started to do that.
And then we work through the programs.
So from an Esri perspective, that's a huge one.
As far as the GeoAI goes,
we have a lot of learning paths.
We also do a lot of documentation on those models
(21:05):
that we stick out there
so that people understand what they were trained
on, how they are used,
and how they're intended to be used such that,
not that they can't use them for anything else.
It's more that here's the starting point,
and you need to understand where you're standing
so that if you want to go from here to there,
you can.
Case in point,
we had somebody take our building rooftop model,
(21:28):
try to run it.
It's a it's based on US data.
I think they tried to run it somewhere
like Saudi
or Asia somewhere,
and it didn't produce great results.
It's because it wasn't trained on roofs
that look like the roofs in that country. Right?
So with a little bit of extra training,
the AEC firms were able to take that
and then leverage it and be very successful with it.
(21:49):
So we try to do a lot of outreach in that regard.
It's also becoming a huge topic
at every one of our events
and every one of our meetings.
(Dave Mulholland)As you know, Micah, we agree with you.
We think all the practitioners
need to have the power of what Esri is doing.
And the GeoAI by putting it on,
we have 2200 employees at VHB.
We went all 2200 to be users of the Esri
(22:09):
software products.
We believe that there's that much power.
We actually have a
a targeted goal to get that on everybody's desk
hopefully in the next year or so.
So moving in the same direction
and I want to actually take
a little detour and launch
because you actually referenced something.
And I agree with you 100%.
This is a transition of, the technology
and the software firms partnering
(22:30):
with the subject matter experts.
So when you think about it from the AEC space,
there are a lot of, technology
based firms
or software based firms
coming into the engineering space.
So, yeah.
Where do you see, Micah,
where do you see that relationship
going over the next five years
relative to the software firms
entering into the AEC space?
(22:51):
And then, Scott, from the flip side,
you represent the,
the engineering firms,
you know, how does that work from your perspective?
(Micah Callough)Yeah, yeah, it's a good one.
I'd say,
and this was definitely my experience
as I had a really good boss
a long time ago, a CIOs said, Micah,
you can't do everything on your own, right?
Build partnerships, build
connectivity, build understanding.
You can accelerate where you're
(23:12):
trying to go by doing that.
And that's exactly what we did.
We built partnerships
with those technology providers.
We didn't see them as threats to us.
We saw them as things that could accelerate
what we're doing.
But but they don't understand the industry.
The engineers brain,
scientists brain, that's not their world.
And so
we use them and work with them in partnership
(23:35):
to make that
such that we're putting what we know
or what you all know as AEC firms
with what they know
in the technology to accelerate
both understanding on each side.
And the other thing,
we are definitely beginning to see, and Scott's
an example of this,
where AECs are beginning
to pull their technology,
true technology people into our space.
(23:57):
I'm a GIS guy.
I was trained as a GIS guy, a degree in GIS,
and then learned all the technology stuff
through practice, right?
So I'm a little slower to the take, right?
Whereas somebody that starts with technology
and goes
the other direction might be a little slower
on the take on the engineering side,
but together we're a really powerful duo
(24:18):
and we can teach each other a lot.
And I have been taught by more technology
people than you can imagine.
That's where I've learned what I know.
(Dave Mulholland)And Scott, from your, from your perspective, you
(Scott Jay Ringle)know.
You got coming in to the, the AEC space
coming from the technology realm, like to your point,
(24:38):
being working from systems integrators, I'm working
for the largest hyperscales
or cloud providers in the world.
Coming from the technology
teams, the deep mind research teams
that come from those facilities.
You're right. It's a technology first company.
But the irony is, I think that
I thought
maybe I'd be bridging or leaving that gap
or moving over to a vertical or a space
where I didn't have the same type of,
(25:02):
synergistic alignment.
And that's and,
sounds like a negative thing,
but it's actually a positive thing.
What I've found
is that being in public sector
and being in AEC is allowing me to bridge the gap.
I used to see it from a client perspective.
VHB would be a client to me.
It's now a very synergistic alignment
on how I can take my various technology skills
(25:22):
and apply them to much more detailed use cases
and get deep into an industry
that I think is incredible.
And screaming for more
AI use cases across the board.
So in a nutshell,
it is a synergistic
blend of companies that are systems
integrators or product manufacturers
in the technology
space, blending with specialists like VHB
(25:42):
to create better outcomes
synergistically together
than that has ever been done before.
Every company today is a technology company.
Like it or not.
Doesn't matter what vertical you're in,
you are a technology company.
You're going to be, you know, a yellow cab versus a
a Uber or a blockbuster versus a Netflix.
So being synergistic together
I think is a great alignment.
(26:03):
I see other competitors
that are trying to do the same thing right now.
And as you said, like other technologists
like myself, coming into spaces in GIS or and in AEC
that are very synergistically aligned
to help drive the business forward.
And, you know, and in summary, I think that's a
the best way I can put it, if that makes sense.
(Dave Mulholland)You know, I just got to see us more actually,
(26:26):
pulling in and partnering more with the software
technology firms.
Yes, VHB made a commitment to bring it in
because, Micah,
we needed to understand
where the technology was going
in order to be educated, smart,
to figure out where
that digital evolution of VHB was going to be.
So is it going to be a takeover?
I don't think it's going to be a takeover.
I think it's going to be a partnership.
(26:47):
I think it's going to be something that
those other firms that we refer to,
the smaller firms
that that we all talk to and target as, partners,
that they're going to need to evolve
and actually become educated
on where they're going as well.
There's one thing that several
the team members brought out to me
is that I understand
that there's like over 70 AI models,
that are trained,
and available for AEC firms.
(27:09):
Maybe if you tell us a little bit about that,
in terms of
what are some of the example use cases and maybe if,
getting into,
you know, where and how does,
how do people access that?
(Micah Callough)So it is in ArcGIS Living Atlas of the World.
You can search that one online.
It's the largest geospatial data
(27:29):
repository in the world.
We spend a lot of money and effort
to make sure that there are cohesive layers,
authoritative information that help the world kind of
be a better place. Ultimately, that's the idea.
And we curate that information
from commercial from public sources,
(27:51):
all kinds of different ways.
And we do a lot of development
for world
wide layers to help the world
think about things like Covid when it spread.
And we were doing a lot of that,
to land use,
to understanding land use across the world.
But then we have a lot of local providers
and even individuals that provide
content through the Living Atlas of the World.
(28:11):
So it's truly,
Jack likes to call systems of systems approach.
The idea is, is with the software,
they can publish the data in an open way,
and everyone can benefit from that information.
And we make that really seamless and easy to do.
We publish a lot of different things
through the Living Atlas of the World.
And one of those things is the deep learning,
(28:34):
toolkit DLTK.
I think if you search for on Living Atlas,
you will see all the models that we put out there.
And those models
range from everything like find the tree canopy
or determine
which trees are here,
or use the segment
anything model to find some tanks in an imagery.
Right?
I can't tell you how many oil and gas tank farms
(28:57):
I had to digitize. At one point in time.
We had to convince them to stop
doing it, in Powerpoint,
I mean, it's just such
inefficiencies in these projects.
And so we were doing it in GIS,
and we thought we were really innovative.
Now with AI, like you talked about earlier, Dave,
that job would be done and done right.
Like I could move from state to state
(29:18):
and get those tank farms,
understand what they're going to be
when you're doing an EIS, when you're starting to
to when one company
wants to buy another,
or when they're responding to an emergency.
We're seeing a lot around building
and footprints and extraction, as a matter of fact,
we've built a layer of buildings
in the Living Atlas of the World,
and it's open source.
It's every building in the whole world.
(29:40):
A little gray box out
shell, which gives you a nice 3D footprint of it.
We used AI to do that ourselves,
so there are just a lot of ways
to pull out information.
There's even a couple in there,
I've seen some recently around, like
looking at if there's a vehicle or a vehicle type
A truck entering a facility or not.
So that
gets beyond the traditional aerial photography
(30:02):
and starts to look at videos.
And we now have the ability to work with video CCTV
feeds, not just imagery.
So you really see this evolving and stepping.
And we put a lot of those models out there
to entice people to recognize that
if they're doing it in a manual way,
here's a better way.
(30:23):
And and I have seen a few projects so far
where a few firms have adopted it
and wiped out the competition pretty quickly.
It is the new way of doing things.
We want to take that complexity
that we were doing before,
and the hours that it took us to do that,
we want to be able to spend those
that time doing something
more creative, more inventive.
We want to move forward a little bit more.
(30:45):
So that's why we stick it out there and we try to, it
helps people to really accelerate what they're doing,
and it's usually their first
taste of AI, if you will.
(Dave Mulholland)Question is, you
know, when they look at that,
when they use it, you know, how do they,
how do they provide feedback to you
and how do you actually get the information
to make the models better?
You know,
because they're there's going to be a lot of users
(31:06):
out there that are going to look at this and go,
wow, there's a lot of exciting tools
that Esri is producing as starting models.
That may want to be a partner
with you to actually evolve.
So how
how do they actually get information
to you or participate to actually,
you know, help these models grow and become smarter?
(Micah Callough)In, oh, so many ways, Dave.
(31:27):
Esri is very, flat, very matrixed organization.
We have things like the, the communities,
the community pages.
There are a lot of communities
that those have direct pipelines
into our product teams.
There's some communities around the
AI and image extraction
and people
that have a lot of discussion
and even help each other out
without Esri even being involved. Right.
(31:49):
There are a lot of channels.
Our product management and product engineering
teams are highly accessible.
If you've worked in or around Esri,
as you've
probably talked
to a product engineer directly more than once,
they're often at our events for communication. Teams
like my team, my solutions engineers.
We are always working through the accounts,
(32:11):
and then we have a much deeper relationship
with companies like VHB at the partnership level,
where we have much deeper conversations
like we did last week around
not only where they think things are at today,
but where things need to evolve tomorrow
to make you successful.
Yeah, I did have some questions around that.
(32:31):
(Scott Jay Ringle)So more specifically, Micah,
I have more on the more technical side
just a little bit.
Anyway,
as far as the models are going, so are there like,
this might be a loaded question, too,
are there maybe 3 or 4 models
that are top, top use cases?
And 70 models is a lot. And are those models?
I'm assuming that some of these are highly specialized.
These smaller models that are tuned
(32:52):
specifically for these exact use cases.
Are those open source?
How locked up
are they?
Can they be trained, fine-tuned
to what levels those kind of things and more
were involved in the actual models themselves.
Are you allowing people to mess with this stuff?
That's what I'm getting at.
(Micah Callough)Very much so.
And we encourage it, right?
That is, the idea is, is that we are
we are putting these out there as a starting point,
(33:12):
not the end point.
And the idea is, is that people will take them
and do a lot of different things.
The big ones
we see now were the obvious ones, people
that we're doing impervious surface
the old school way are now doing it
through extraction.
People that want to extract assets,
like we've seen a lot around the power poles
and the sag and the power lines and things like that.
People that want to build
more realistic scenes with the trees
(33:34):
and stuff like that in 3D,
the Segment Anything Model has definitely been one
because there's a lot around
not knowing what you're hunting for
and just kind of looking for it,
parking spaces and things like that.
But we
absolutely encourage people
to take them, use them and adapt them.
I've seen others that, and even
including myself,
(33:55):
where I've taken them
and adapted them for doing things like extracting,
assets along train corridors, which are,
you're very difficult to get into.
You have to be Rail Safe certified
and using two cameras,
that are GoPros and a GPS point
and the snapping off pictures
every couple of minutes,
which means we can start to extract that data
(34:16):
in a way
that was previously more heavy
and hard to do with LiDAR data.
A lot easier to do with imagery, and we can use LiDAR
data to back up and check our work if we need to.
When we need to up the probability.
So we just see a lot of people
adapting them every day.
I walk into a new customers. Why?
I like to go sit with customers like VHB
(34:38):
because I learned a lot being there and listening
to that, the team and what they're doing,
and there's just some amazing innovations.
It's what I hoped would happen
five years ago is really, truly beginning to happen
is as we focus more on AEC,
they're starting to jump and step faster and faster.
It's amazing.
(Scott Jay Ringle)It is amazing.
(34:58):
And that's really good to hear
that type of collaboration
you're allowing to happen
with these with these models.
Because that's what it takes.
And coming from,
as everyone knows,
like coming from Microsoft as an example,
being able to use open source models
and obviously closed source
models are a big foundational piece of Microsoft,
don't get me wrong.
But the bottom line is,
open source,
smaller models are becoming the way that we see
(35:19):
AI be deployed on a more common,
on a day to day basis.
It's less expensive, it's easier to tune.
You have a better outcome with the result.
It is, the 70 models you have on the allow
the and the ability
to collaborate very closely with the team
and go out and use it and customize it,
I think is a huge step forward
in getting the adoption of AI across,
all the verticals we're talking about.
(35:40):
It's incredibly powerful, actually.
So I'm glad to hear that.
(Micah Callough)Yeah, we're
we're if nothing else,
the Living Atlas, we put a lot of effort into.
The other thing that we do
a lot of
is the open standards groups,
OGC, buildingSMART International.
A lot of the open things,
we don't just say we're open, we put a lot of money
and effort
and we publish open standards for everyone to use.
(36:02):
We're not just about what we do with it.
It's it's literally,
the building scene layer is a good example of that.
I3S, that is an open
standard that anybody in the world can use.
That's excellent.
(Scott Jay Ringle)I love to hear it.
And back to you, Dave.
(Dave Mulholland)So, Micah, I,
wanted to go back to one of the previous comments
(36:23):
we were talking about is really AI,
AEC space, the ethical, conundrum that I think,
that AI could present if not, careful, sensitive,
PII data that's out there.
What conversations should firms be
having around this relative to it
within their organizations?
(36:44):
And what is Esri doing to actually
also helping to actually get that message out?
(Micah Callough)Yeah.
First and foremost, absolutely.
This should be at the center of any conversations.
The security, the privacy information.
I worked for a global firm, so maybe the US
not so far ahead in this is you would see.
(37:06):
We're at a state now where all this is going on.
There's going to be some
there's going to be an injection
at some point in time at the government level.
Right. Like it's inevitable.
It's going to have to be.
It's our responsibilities to understand that.
But AEC firms have an added extra responsibility
because an AEC firm puts a stamp on something
(37:27):
and says,
this is what we're doing, and I'm an engineer
and this is possible to do. Right.
And not all of our work
is about building something and stamping it.
But the reality is a lot of it is,
and or a lot of it comes with the expectation
that an engineer said, I can do this.
Therefore it's something I can do.
So if we're relying on AI that's rolling on bad data
(37:49):
and making decisions
about how I should engineer something
that's not how AI should
or could be used right now, in my opinion.
Now, could it help the engineer
to make some decisions
and think through some iterations of what
they're trying to do
so such that they understand
what that decision was being made from? Absolutely.
So that's my biggest like.
(38:10):
And I get it all the time
because I want to run forward with the technology.
And then the engineers like,
you know, there's
this level of kind of freak out for a moment
and then we all calm down for a minute
and we start talking it through and understanding
what we're doing and what we're using it for.
And we go, well,
you know, this could actually help me.
So you've got to get through that moment of chaos,
(38:31):
I think, in these situations
and then work through, like Scott said, change.
(Dave Mulholland)I again, I did mention
that, you know,
I've been in the business 30 years
and I sit there and look at the parity
between when I started and where it is now.
And there's a lot of fear out there relative to AI.
And I go back to when I started,
traffic engineer by trade,
you know, I ended up having to do a strain
(38:54):
pull calculation for a signal by hand.
I did a calculation by hand,
and then all of a sudden
they handed me a software package that said,
can you do it using this?
It was fractions of the time.
So I relate AI as being very similar to that.
Right.
You're actually having another tool
that you need to actually look at.
Check, validate and make sure that you,
(39:15):
do not get those hallucinogens
or errors that come out of the models.
But realistically, it's incumbent
on us, the engineers in the room,
to actually make sure that happens.
(Micah Callough)I mean,
the other thing, Scott, I'm sure you see this is
some of the reactions
is to stick the head in the sand
and kind of try to ignore it. Absolutely.
(Scott Jay Ringle)That was one of the blockers.
(39:36):
We were talking about that just yesterday, in fact,
not taking action or not paying attention
or thinking this is another fad
or just not wanting to adjust and change.
Absolutely. See it a lot.
That and data.
That's one
and two blockers
right there for implementation of AI.
(Dave Mulholland)So I was going to take the step forward now.
(39:56):
We mentioned there's three different areas.
I want to start to look forward into,
the next five years,
in terms of some of the significant advancement.
But so, Micah, I'm coming to you Esri, GeoAI, Launch,
where are we going to go over the next five years?
And where do you see it going?
And what do you hope to,
you know,
get accomplished over
the next five years in the AI space?
(40:18):
(Micah Callough)Yeah.
So I'll, I talked a lot about GeoAI because that's
the immediate low hanging fruit,
the thing that people can wrap
their hands around right now.
There are also smaller things that are happening.
And I can't say that they're smaller.
They're just more behind the
scenes and more ingrained.
One, for example,
is we were building a lot of AI into our systems.
(40:39):
I talked about that idea
of weaving stuff in AI into the product.
Survey123 on the,
I can't remember what version of,
is the latest beta at this stage.
Or it might be the latest
production release, has an AI function
that I used a couple days ago.
I asked it to create me a form for a webinar
or a meeting
I was going to have, where I wanted feedback on
(41:00):
blah blah
and I fed it some information
and boom, it created me a form
that I could send out to do a survey on.
So that would have taken me
an hour or two to think and come up with.
Did it, was it perfect? No.
There's a couple things I threw out of there
and some stuff I added
because of my knowledge
and what I'm seeking
as far as information
that I didn't prompt
the AI for, I didn't feed it enough good information.
(41:23):
But man, it did get me pretty far pretty fast.
And we're really thinking
about that across all of our products.
Back at the Federal User Conference
earlier this year,
some of you may have seen a, an AI prompt tool
that was designed to allow people to ask questions
that are geospatial
and not have to go to a GIS person to get the answer.
(41:43):
So, you know,
I want to understand my property,
you know, what's the value of it?
Where you know what, what's within walking
distance of this area?
Is this property commercially viable, right?
Just being able to do things
and get to geospatial information
without knowing
what geospatial is in the first place.
So those are a lot of the assistant-type models.
(42:05):
And then underneath the hood,
we've talked a lot about people
who build applications.
And we have to know to turn this layer on
and this thing and click this tool
and click this button.
An engineer would probably just rather
ask it a question and get an answer, right.
They don't want to know about those buttons and tools
and all the things that are going on.
So how do we make take the AI
and the assistant type chat models
(42:27):
and make that better and faster?
So that's a little bit
of where Esri is focusing its attention.
There's a lot of deeper R&D things
that are, that are getting really interesting.
And we're just kind of
in that balancing game of like, where are the models?
Is GPT the thing and always going to be the thing?
I'm not so sure it will be, right, long term.
(42:47):
Scott can help me with that.
(Scott Jay Ringle)I think AI will completely transform
how we design and build
by enabling real time
dynamic project design and execution.
And that sounds like a marketing term,
but I want to, when I mean that,
I mean that sincerely.
I see a future
where AI is integrated into every phase of a project,
from concept to construction.
(43:08):
It's basically
AI-driven systems will continuously analyze data,
adjust designs,
and even make real-time decisions on site.
So imagine this, if you will,
a condition changes on a construction site,
a construction site, whether it's weather,
supply chain disruption, unforeseen challenges,
(43:30):
AI algorithms will automatically adapt project
plans, reroute resources and adjust timelines.
It will continue with the overall goal
of the project timeline
and keep all this in mind
at the same time, simultaneously, which is a,
this will eliminate
many of the delays and inefficiencies
that plague the industry.
A huge piece of where this eventually is
(43:50):
headed to where there's
so much automation, there's people involved.
It's a copilot,
if you will, not to overuse that word,
but it's a copilot for the people
that are involved in these projects
at all aspects and everyone
co-developing this together.
But the AI streamlines the process to that level.
Architects, engineers,
contractors will all be alongside
with your AI assistant, allowing the the creativity
(44:13):
and the efficiency to be enhanced ten fold,
100 fold.
In short,
if I was to summarize this,
the future will be a blend of human expertise
and any AI-driven precision
to create a more responsive process for design
across the board and architecture.
I mean, that kind of philosophy
is where I'm going with this, if that makes sense.
(Micah Callough)I 100% agree.
(44:33):
I saw that,
I was at a construction event recently, and I
and they had an
AI as one of the topics during the afternoon
they were talking about.
And I kind of thought, well,
this will be interesting, right?
Like these guys aren't using it
and they were quite well using it for like
figuring out billing issues, looking at risks stuff.
They were very much adapting about what
they were doing,
which was really good, was choking it
(44:55):
down to the problem that they have.
So constraining it for risk.
But I find it interesting that it's, it is it's the
this is the fastest I've ever seen
anything get adopted in this industry right?
Like ever.
And more people are excited about it
that I, that I think barely can open up
Excel or Word. Right.
(Scott Jay Ringle)That's so true.
(45:16):
(Dave Mulholland)I do want to pivot and actually go
with a closing question.
And this is a, this is a general one.
One takeaway, right.
If you had to advise a firm that's just now
getting into AI, that's just now starting to use AI,
what is that one action that they should take today
(45:36):
to actually prepare
and get going into the AI space specifically?
And you know, Micah,
it can be using it, the GeoAI
specifically if you like.
(Micah Callough)Well,
you know,
as much as I'd love
to sell GeoAI and ArcGIS, I'm actually going to say
it's a little bit more about
(45:57):
bringing, it's the people, right?
It's going to be about the people, right?
I get it's a lot of people management,
a lot of change management.
Getting the right partners,
getting the right staff brought in
that has the experience that can help
accelerate the rest of your staff,
helping the people
to understand where it's going to be.
Technology is usually never the problem.
(46:18):
It's usually the people transformation.
And I know that from experience,
I have been beat up
more than my fair share
trying to implement technology inside of an AEC firm.
But I think now is the right time
and there are people out there that are really,
they love to solve these problems, right?
So work with them,
figure out
how you partner with them,
(46:39):
and by all means leverage GeoAI if you want to try
to start doing something, right, like that's
a good starting point for some of those folks.
(Dave Mulholland)So, so as
as the engineer in the room, I agree with you.
I think GeoAI is actually one of the first entry ways
and easy
opportunities for a firm
to actually start to engage and embrace.
(47:00):
And I'm looking forward to, to the opportunity
to actually
use it on
some of the activities
that we do, across the entire footprint.
Scott, the one takeaway,
what's that one action
that you have to basically advise the listeners to,
to, to take note of.
(Scott Jay Ringle)Don't be afraid to move forward.
And it comes down to a single small little word:
(47:22):
data.
Talk to us about data.
Don't be afraid to have your data prepared and ready
to move forward.
And again, that's a general statement.
But the the net net of this is
the single biggest blocker, and to Micah's point is
that people,
the people are already in the mindset
of being ready to move forward with this technology.
(47:44):
The number two blocker is going to be
their data assets and their data state.
So we need to talk about that
and demystify that complex data that they have,
which is not as hard as it sounds.
There's a lot of fear,
uncertainty and doubt,
as they say in sales terminology,
wrapped around data assets.
It just takes someone that knows how to analyze that,
come in, have a conversation, do a discovery workshop
(48:06):
to get you ready
and prepared to go down that journey to build out AI
applications and platforms at scale,
because it all starts there.
From there, it becomes magical.
That's no pun intended,
but that's that's my takeaway, Dave.
It's going to be analyze
data and data data data it all starts there.
(Dave Mulholland)You cannot do AI without good data.
(48:28):
And I think, Micah, you referred to... Good data.
You refer to that.
(Scott Jay Ringle)Yeah. Good data. I apologize.
Yeah. You know good, good data there.
I was sitting there and I was reflecting
and it's one of those general takeaways.
I think it's the eyes wide open, right?
Relative to where AI is in the industry is,
I was going to come back with a, you know,
my response was the policies
and be aware of what's going on within it.
(48:48):
Micah,
you referred to the state agencies
as some of the regulations that are coming.
I think for folks
that, would be listening
to this, one of those key messages
that we got out is
agencies are already starting
to look at this, already
starting to understand the impacts of it.
When you think about procurement
and going after a project, oftentimes
now they're including some provisions
(49:08):
relative to the use of AI in your pursuit.
Be careful.
And some of these are saying you can be disqualified
if you use AI as part of your submission.
Same thing on the project delivery.
When you think about like scoping a project
and scoping
and delivering a project for a client,
some of the state agencies across
the US are now looking at if you do not inform
(49:31):
the PM that you're using AI as a delivery,
then you could potentially be
disqualified from actually this submission.
So it's a it's an eyes wide open for me.
It is here.
It is moving.
I think both of you indicate
it is moving exponentially.
Micah, I'll be honest with you, I'm
looking forward to seeing the excitement of where
Esri is taking this over the next 5 or 10 years,
(49:53):
when I actually get my next 30 years in
in the industry.
(Micah Callough)There's a lot of
deeper R&D things
that are getting really interesting,
and we're just kind of
in that balancing game of like, where are the models?
Is GPT the thing and always going to be the thing?
I'm not so sure it will be right long term.
Scott can help me with that.
(Dave Mulholland)And I think we're hitting
probably one of the biggest,
one of the challenges
or attempting to do that
with this podcast, right, is education
(00:26):
and training of what AI is and what AI isn't.
(Scott Jay Ringle)Don't be afraid to move forward.
And it comes down to a single,
small little word (00:33):
data.
Talk to us about data.
Don't be afraid to have your data repaired
and ready to move forward.
(Dave Mulholland)Thank you for joining us today.
We're excited to come today to talk
about artificial intelligence.
Artificial intelligence
AI continues to be the one of the most influential
(00:54):
yet disrupting technologies
to impact the AEC industry in decades.
In some cases, we're seeing that is impacting us
somewhere between 25% up to even 95%,
in terms of the automation of where we're at.
ChatGPT engulfed
the industry in November 2022, and here we sit
almost two years later
(01:16):
seeing the impacts of other technologies,
and other tools that are moving forward.
The next ten years will completely be revolutionized
with the impacts of AI.
The industry needs to figure out
how to harness this technology, use it,
innovate, and automate
where we're moving as part of the AI revolution.
I am really excited to be here
(01:36):
co-hosting this with Scott Ringle, VHB's very own
AI lead.
Scott comes to us with a background
in the technology space, which complements
what we're trying to do with VHB.
He brings a wealth of experience, actually driving AI
and the practice in the business,
and we're going to hear a lot from him.
Scott, thank you very much for joining us today.
(Scott Jay Ringle)Dave, Micah, it's great to be here.
(01:56):
Thanks for having me on.
Looking forward to our conversation.
(Dave Mulholland)So today Scott and I are joined by Micah Callough,
Technical Director for AEC and Environmental
Consulting with Esri.
Micah,
we're extremely grateful for you
to take the time and join us today
and talk about this very instrumental
and challenging topic.
(Micah Callough)Yeah, sure. I'm glad to be here with you guys.
(02:17):
My name is Micah Callough,
I'm the Technical Director
for AEC and Environmental Consulting here at Esri.
My focus is on helping connect
kind of the AEC business
with,
what Esri does and kind of think about it
from a strategic perspective.
I joined Esri maybe five years ago now, seems,
(02:38):
gone fast.
Right before Covid and, I've helped to grow the
the sector
and the focus on AEC here at Esri in that time.
Prior to that, I had worked,
30 plus
years in the AEC business
between two different companies.
So I've spent most of my life
as a practitioner
on the AEC industry side in many different areas,
(03:00):
all over the place for all the things that AECs do.
Hey, Scott, thank you for joining us.
And why don't you tell us a little bit
about yourself? Sure.
(Scott Jay Ringle)I'll give you a quick summary here.
My journey began back in 1992 in IT,
came a long way since then.
But passion,
my passion
for AI formally began in 2017,
which I think is relevant to our conversation today.
(03:21):
After reading the now famous whitepaper
on transformer models
called Attention is All You Need.
Kind of the rest was history
after that, I was obsessed with the technology
and proceeded to hone
my skills at companies like DXC Technology,
one of the largest global systems
out there, and Google and Microsoft,
as you already mentioned.
I kind of specialize in foundation
(03:42):
model architecture and solving problems
for our fortune 500 clients
over at those locations where I was at before.
Now specializing in machine learning frameworks,
including neural networks,
and I helped drive advanced AI use cases here at VHB,
focusing on all things
AI and helping integrate it
into all of our workflows
and drive innovation
and deliver better outcomes across the board.
(04:03):
That's why I'm here.
(Dave Mulholland)Thanks, Scott, I appreciate that.
And,
I really wanted to take this session and really
break it down into like three different areas.
First area I wanted to talk about:
What is AI doing in the industry?
So this is a high level
what AI is actually in terms of disruptors,
some of the benefits,
some of the challenges that we're seeing.
The secondary that I wanted to get into
(04:24):
is really, Micah, this gets to you
and what Esri is doing.
Where is Esri going?
You know, what are we working on?
And then the last part I wanted to go into is,
Where do we see it going?
This is a challenge.
And I don't know that we actually fully
have our arms wrapped around it.
Three years, maybe five, ten, 15 years.
There's going to be some challenges.
Everybody in the industry is going to be facing it.
So let's start out with some basics.
(04:45):
Scott, I'm going to come to you first
and then Micah, I'll let you go next.
Where do we see...and for
the listeners that are new to AI...What
is the definition of AI?
What does it mean to you? Okay.
(Scott Jay Ringle)Let me give you my personal opinion on this first.
So to me personally,
artificial intelligence in this case
focusing on the extreme is a game changer
(05:08):
that enhances how we design,
engineer and build across the board.
Basically, it's
about bringing smart adaptive technology
into our workflows daily
and enhancing our capabilities.
So that's my personal kind of umbrella around this.
For my experience
AI simulates
human intelligence processes like learning,
reasoning and problem
solving through computer systems.
I mean, that's what I really is.
(05:30):
I've seen firsthand how I can shift
through massive amounts of data
to help us make better informed decisions,
anticipate project outcomes,
even suggest design alternatives, etc.
specific to AEC
so consider it your tireless assistant,
if you will, that handles the heavy data
lifting and allowing us to focus more on
creativity and strategic problem solving.
Kind of my philosophical view of AI.
(05:51):
Well, there you go.
(Dave Mulholland)Thanks, Scott and Micah. I did this on purpose.
I was like trying to get the general.
And then I know
Esri as we play in the geospatial world, what is Esri
and what is your definition of AI
and what does it mean to the listeners and
where we're going?
(Micah Callough)Well, when I think about it
and I cheat because I'm very much an, a, an AEC
practitioner, right. That's where I've lived my life.
(06:14):
My job has always been connecting the technology
and the business needs. Right.
Like really try to understand it.
And AI is not, is nothing new.
It's been around and components of it
have been around for quite some time.
It's just now starting to get that
the technology,
the demand and the drive is beginning
to start to match each other.
And coming over here on the products side
(06:36):
or really the drive for me
there was that
I think the AEC industry is sitting at a spot
where it can, it's
going to start to shift in a really big way.
And I could see that coming from the seat
I sat in before,
and I felt like coming over on
the product
teams would start to really help,
like, could start to help,
to steer that a little bit.
Esri has played in this space quite a bit.
(07:00):
And the thing that's most
like visible,
I think, to our traditional geospatial people,
is what we refer to as GeoAI,
which is,
quite frankly, machine learning,
image extraction, getting details out of it.
And that's
where a lot of people start their AI journey
that aren't practitioners of IT,
and technologies
that are deep down in the bowels of the technology,
(07:21):
but the business people,
and that's where they see a lot of value, right?
We do a lot in geospatial
with imagery and image stuff.
So that's where Esri really got
its start and its positioning.
But like all other technology
companies
that developed platforms like Esri,
including companies like Autodesk and Esri,
who are partnered together,
we're all really beginning, just like you all,
(07:43):
to start to think about how AI, is it a thing,
a monolithic thing,
but how it's something that starts to get woven into
what we do to accellerate or to improve upon it?
(Dave Mulholland)I agree with you hundred percent.
I think we're to the point where you're starting
to see this upward movement of where AI is.
Eventually, it'll be part of the daily practice
(08:04):
and who we are
and what we do as engineers, planners, and scientists.
So,
one of the questions
I get asked a lot, right,
and these are by people that do not know
or do not understand
AI, is
what are some of the most significant benefits
or areas that you're starting to see movement in,
using AI within the AEC space?
So, you know, Micah, I'll come to you first.
You know,
where do you see some of the biggest benefits
(08:25):
in terms of return on investment or,
or how people are using AI?
(Micah Callough)Right now,
the one that's actually truly in the business
that I see is the GeoAI stuff.
It's those models
that we're
beginning to put out
on the Living Atlas of the World,
that we create and train on different imagery.
We put publish them out into the world.
(08:45):
They're free to pick up for anybody
that's got an ArcGIS license,
and they can start to run them.
And those have been traditionally
being run
in our professional desktop tool, ArcGIS Pro,
kind of targeting that more data
engineer, data analytics
kind of persona and personality,
where we're even starting to see those tools
start to become able to access
(09:06):
within the online environment.
So your casual web
based users
can start to get information off of that imagery.
And why I like this the most is,
is that the AEC industry,
or really the geospatial side of things,
has relied on imagery to,
to be a major component of mapping,
basically making your road map,
(09:27):
making your street map, making a parcel map,
making maps of all the buildings.
These are all have always traditionally
been photogrammetry,
image-based type things and or survey.
And we always want to get imagery
on our projects, right?
But we could, in the old days
it was kind of hard and expensive to do that.
So we didn't always get the best imagery.
(09:49):
Well, now imagery and the technology behind
it is becoming more accessible
and readily available through drones
and through other platforms and commercial offerings
that all of a sudden
we've got a lot of imagery
and observations of what's
going on out in the real world
that we want to take advantage of,
and we can take advantage of it
in so many different ways because of AI.
(10:11):
And what why we coined at GeoAI
is because we're looking at
and extracting geographic location
meaning from those images.
Everything from, is there a car here?
Where's the fire hydrant?
Get me all the trees.
Stuff that we can do en masse
and for a lot less money, right?
Which makes project managers
a whole lot more happy right now.
(10:32):
We can do things better, faster, cheaper.
That's really the
the starting point
and the one that I see most practically being used.
That's not an experimental stage.
It's, it's truly in a production stage of use.
(Dave Mulholland)You know, Micah, I've been in the business 30 years.
I wish I would have had this tool 25 years ago
to help with my career.
So if we go back in time,
if we can go back in time and actually do that,
(10:54):
that would be amazing.
Scott, turning to you
now in terms of the benefits, where
where does
VHB see
some of the largest benefits or movement of AI?
(Scott Jay Ringle)Well, Micah,
his answer was extremely detailed and good
and mine
is going to be a little bit
more generic wrapped around
design services for VHB
and where we're really focusing right now
it is the biggest impact
(11:15):
that we see an AI currently
using within design services space.
So AI can generate multiple design iterations
and optimize for factors like sustainability
and cost,
and predict
how designs will perform under various conditions.
Those are major leaps forward also,
and how we can enhance productivity cost,
go to market, everything that's wrapped around,
design services,
(11:35):
AI has radically enhanced that capability.
It not only accelerates design process,
but also leads to more informative sand effective decisions.
So again, decision making across the board.
Everything has been enhanced through AI,
adopting AI driven tools and methodologies.
This design industry we were talking about
is to use greater efficiency
through various technologies.
We'll get into more later in this call.
(11:56):
But bottom line,
the impact that we see
the most benefit from
is specific
to the verticals
that we're in, is in the design process space,
if that makes sense.
(Dave Mulholland)And I don't know that you'd have any disagreement.
And Micah was actually referring
to their, their partners over at Autodesk.
And I know, you know,
when you think about the geospatial world
that Esri plays in and the design services side,
(12:16):
it does go hand in glove.
And I know that's a lot of effort
that Esri and Autodesk
and the other vendors I work with as well.
Yeah. So we talked to go ahead go.
(Scott Jay Ringle)No go ahead I was going to say ultimately leading
we're basically getting to better design buildings
and infrastructure, city automation,
AI processes and design functionalities we have.
I mean, that's what it comes down to.
Better design, better
buildings, better infrastructure based on AI.
(12:37):
(Micah Callough)So many more options, right.
Like you can do
and think about it
in so many more ways with all those crazy
overlays of sustainability and resiliency
that we have dealing with hurricanes
like we just had the other day,
if we need to consider these things in design
because it's our new normal.
(Dave Mulholland)Yeah, we're starting to see a lot of that movement.
You talk about the sustainability and the resiliency
(12:58):
and moving in towards the design services.
You know, Scott talked about design,
the embedded carbon calculations
and how we're doing it, how do you optimize it.
Scenario planning.
What are some of the,
what is you know, Micah,
what are some of your AEC clients
or some of the challenges
you have with implementing the AI?
What's keeping them up at night
and how do they actually stay ahead
(13:19):
of some of the competition driving this?
(Micah Callough)Yeah, the AI
term is a double edged sword in a lot of ways, right.
Because it's now the hot thing.
We are we still are talking about digital twins.
And that was a hot thing for a minute.
And for that it was big data.
And there's always a hot thing.
And I'm very happy to see that the C-Level in AEC
(13:40):
firms are beginning
to understand that business
transformation happens with technology, not,
outside of the technology.
Right? Like so it's a component of it.
So there's a lot going on there
that's driving that forward.
Some of the risks that I see when I look across
many of the organizations is in general,
(14:01):
it's the data? Like their data,
the most valuable thing an AEC has
is its knowledge and its data that that knowledge creates.
And right now most of that is created in files here.
And files,
there in somebody's
computer over there
and some SharePoint drive over there.
(14:22):
And
it's not, it's thought of project by project
because that's
the project managers drive, do
the project deliver the project, right?
But that data has a lot of value
that people don't even realize at this point in time.
And so if if the data is not organized or accessible
or something
that you can actually take advantage
(14:42):
of to interrogate with AI,
then you then have to start there first
to then be able to take the AI step.
So it's,
there's no like, magic
silver bullet
where we're just going to feed it
random stuff,
and it's going to give us the amazing answers.
We're going to have to do some data
engineering work to make that work the right way.
(Dave Mulholland)Yeah, yeah.
Scott, I know, I know,
(15:03):
data has been big for you and where you're going.
I think Micah took your thunder
there a little bit by saying data, because that's it.
Where do you see
some of the biggest challenges with the AI side?
(Scott Jay Ringle)The biggest challenge is on the data side.
So biggest challenges with AI? I actually have 5
things I'd like to say about that.
Number 1 as Micah already pulled it out
(15:23):
or pointed out data quality and availability.
That's a huge one.
AI models are only as good as the data
they're trained.
So inconsistent, incomplete, or biased
data can lead to inaccurate predictions and insights.
I mean, it's not just in this industry,
in this vertical,
it's in all of them
coming from the systems integration
world of working
with many different verticals and markets.
We see this is a common
denominator across
almost every aspect in building out AI models.
(15:46):
Integration with existing systems is another problem.
So when you incorporate AI to establish workflows
and I talk to workflows,
and legacy systems can be technically complex
and resource intensive.
Having the right teams that know how to do that,
these are blockers to people
that are trying to get into AI
in a more advanced level,
and there's ethical and legal considerations.
(16:06):
The other one I think is important is a skills gap.
Again, these are coming more
from the systems
integration side
of seeing the general common denominator, core issues
when it comes to challenges
with AI and integrating it into existing
or even new start up systems.
Skills gaps is one, another one is change management.
Overcoming organizational resistance
to new technologies.
This is not just this.
(16:27):
This could be again,
as Micah has said, there's always a new thing.
I remember to date myself
a little back in the day when the, the data people
and the
IT infrastructure teams
and the voice teams didn't get along and they started
merging voice over IP and infrastructure.
That was fun. Same kind of scenario.
So change management is an issue.
Those are my five concerns
for biggest challenges in AI.
And that's, some pretty common denominators.
(16:48):
Like I said.
(Dave Mulholland)Yeah.
And I think we're hitting
probably one of the biggest,
one of the
challenges are
attempting to do that with this podcast,
right, is education and training
and what AI is and what AI isn't?
Today I think AI actually is that great equalizer.
Firms, smaller firms have the ability
to leverage and partner with Micah, firms like Esri
(17:10):
and actually take in and really,
work with their data, actually
aggregate it,
put it into a right, structure format
and actually moving forward.
So how, you know Micah
Esri is one of the biggest providers
of geospatial in the country, if not globally.
How does
Esri do that training program
to get the message out of what you guys are doing?
(17:30):
And this is more of a
to give you a platform to say, here's where we are
and here's where we're going
relative to the training side of it.
And then specifically getting into your GeoAI.
And the roll out
where people are going to be using that.
(Micah Callough)Yeah, I mean,
Esri has a long history of having to
build its own people, so to speak.
(17:51):
When GIS first came on the scene
a little over 50 years ago,
it was not something
that was in the in the masses, right.
So Jack Dangermond, owner of Esri,
did a lot of work in building GIS professionals
within schools and education
and drilling back into those things
(18:12):
and seeding that out
so that people kind of understand what geospatial is,
not as in a software,
but as in it's a science, right?
The idea of mapping the world is a thing.
So there's a lot of need to try to push back in
to the education systems
and make sure that what we're churning out into
(18:33):
the industry
is, is going to be people that understand
the beauty of that is, is most of these kids
have been working with this stuff
more than any of us have. Us old people.
And they come to it
naturally in some cases,
I'm finding that you can pick up people that know
a lot about it at a very young age
and become very useful and effective in effect,
(18:55):
they're doing exactly what you just said, Dave.
They're using this new tool and this new capability
to accelerate
what they're doing in their career path.
And a lot of us are standing around
going, wow, that's amazing.
Initially, it's
going to look like voodoo magic, right?
But at the same time,
we have to take your risk question from before.
(19:16):
We still have to take it with a grain of salt. Right?
Like we need to make sure
that what we're getting is true information,
that it's actually matching what we're doing.
Esri does
also a lot of effort, as you know, at conferences,
at workshops we put out learning paths.
We do everything and anything we can to try to help
(19:37):
people understand
what they're doing
so that they're not opening up some software
and clicking the next button,
getting a result and not understanding
where that came from.
Which is why we say that
when we talk about geospatial professionals
getting into this area of AI,
we start to
see the developers
that we've always had kind of using geospatial.
(19:58):
And now we're beginning to see this analytics
and data engineering and data science community
that are being developed.
And where
we're starting to show
is that they need to be thinking
about how they work together.
And so we're trying to do a lot of change
management or evolution.
There's another name Esri likes to talk about.
I'm just used to change management.
But trying to help those teams come together
(20:20):
because it's not about like
taking a bunch of geospatial people and saying,
you're all going to do this now.
It's more about merging them together.
Case in point,
when we got into doing water
models and utility models back in the late 90s,
and the geometric network
came out in the 2000s,
and we were starting to model out
all these utilities.
(20:42):
We didn't do that alone with GIS people alone.
We did that with the engineers and the
the people that know the water utility
providers, right?
The people that know that business.
And together they started to do that.
And then we work through the programs.
So from an Esri perspective, that's a huge one.
As far as the GeoAI goes,
we have a lot of learning paths.
We also do a lot of documentation on those models
(21:05):
that we stick out there
so that people understand what they were trained
on, how they are used,
and how they're intended to be used such that,
not that they can't use them for anything else.
It's more that here's the starting point,
and you need to understand where you're standing
so that if you want to go from here to there,
you can.
Case in point,
we had somebody take our building rooftop model,
(21:28):
try to run it.
It's a it's based on US data.
I think they tried to run it somewhere
like Saudi
or Asia somewhere,
and it didn't produce great results.
It's because it wasn't trained on roofs
that look like the roofs in that country. Right?
So with a little bit of extra training,
the AEC firms were able to take that
and then leverage it and be very successful with it.
(21:49):
So we try to do a lot of outreach in that regard.
It's also becoming a huge topic
at every one of our events
and every one of our meetings.
(Dave Mulholland)As you know, Micah, we agree with you.
We think all the practitioners
need to have the power of what Esri is doing.
And the GeoAI by putting it on,
we have 2200 employees at VHB.
We went all 2200 to be users of the Esri
(22:09):
software products.
We believe that there's that much power.
We actually have a
a targeted goal to get that on everybody's desk
hopefully in the next year or so.
So moving in the same direction
and I want to actually take
a little detour and launch
because you actually referenced something.
And I agree with you 100%.
This is a transition of, the technology
and the software firms partnering
(22:30):
with the subject matter experts.
So when you think about it from the AEC space,
there are a lot of, technology
based firms
or software based firms
coming into the engineering space.
So, yeah.
Where do you see, Micah,
where do you see that relationship
going over the next five years
relative to the software firms
entering into the AEC space?
(22:51):
And then, Scott, from the flip side,
you represent the,
the engineering firms,
you know, how does that work from your perspective?
(Micah Callough)Yeah, yeah, it's a good one.
I'd say,
and this was definitely my experience
as I had a really good boss
a long time ago, a CIOs said, Micah,
you can't do everything on your own, right?
Build partnerships, build
connectivity, build understanding.
You can accelerate where you're
(23:12):
trying to go by doing that.
And that's exactly what we did.
We built partnerships
with those technology providers.
We didn't see them as threats to us.
We saw them as things that could accelerate
what we're doing.
But but they don't understand the industry.
The engineers brain,
scientists brain, that's not their world.
And so
we use them and work with them in partnership
(23:35):
to make that
such that we're putting what we know
or what you all know as AEC firms
with what they know
in the technology to accelerate
both understanding on each side.
And the other thing,
we are definitely beginning to see, and Scott's
an example of this,
where AECs are beginning
to pull their technology,
true technology people into our space.
(23:57):
I'm a GIS guy.
I was trained as a GIS guy, a degree in GIS,
and then learned all the technology stuff
through practice, right?
So I'm a little slower to the take, right?
Whereas somebody that starts with technology
and goes
the other direction might be a little slower
on the take on the engineering side,
but together we're a really powerful duo
(24:18):
and we can teach each other a lot.
And I have been taught by more technology
people than you can imagine.
That's where I've learned what I know.
(Dave Mulholland)And Scott, from your, from your perspective, you
(Scott Jay Ringle)know.
You got coming in to the, the AEC space
coming from the technology realm, like to your point,
(24:38):
being working from systems integrators, I'm working
for the largest hyperscales
or cloud providers in the world.
Coming from the technology
teams, the deep mind research teams
that come from those facilities.
You're right. It's a technology first company.
But the irony is, I think that
I thought
maybe I'd be bridging or leaving that gap
or moving over to a vertical or a space
where I didn't have the same type of,
(25:02):
synergistic alignment.
And that's and,
sounds like a negative thing,
but it's actually a positive thing.
What I've found
is that being in public sector
and being in AEC is allowing me to bridge the gap.
I used to see it from a client perspective.
VHB would be a client to me.
It's now a very synergistic alignment
on how I can take my various technology skills
(25:22):
and apply them to much more detailed use cases
and get deep into an industry
that I think is incredible.
And screaming for more
AI use cases across the board.
So in a nutshell,
it is a synergistic
blend of companies that are systems
integrators or product manufacturers
in the technology
space, blending with specialists like VHB
(25:42):
to create better outcomes
synergistically together
than that has ever been done before.
Every company today is a technology company.
Like it or not.
Doesn't matter what vertical you're in,
you are a technology company.
You're going to be, you know, a yellow cab versus a
a Uber or a blockbuster versus a Netflix.
So being synergistic together
I think is a great alignment.
(26:03):
I see other competitors
that are trying to do the same thing right now.
And as you said, like other technologists
like myself, coming into spaces in GIS or and in AEC
that are very synergistically aligned
to help drive the business forward.
And, you know, and in summary, I think that's a
the best way I can put it, if that makes sense.
(Dave Mulholland)You know, I just got to see us more actually,
(26:26):
pulling in and partnering more with the software
technology firms.
Yes, VHB made a commitment to bring it in
because, Micah,
we needed to understand
where the technology was going
in order to be educated, smart,
to figure out where
that digital evolution of VHB was going to be.
So is it going to be a takeover?
I don't think it's going to be a takeover.
I think it's going to be a partnership.
(26:47):
I think it's going to be something that
those other firms that we refer to,
the smaller firms
that that we all talk to and target as, partners,
that they're going to need to evolve
and actually become educated
on where they're going as well.
There's one thing that several
the team members brought out to me
is that I understand
that there's like over 70 AI models,
that are trained,
and available for AEC firms.
(27:09):
Maybe if you tell us a little bit about that,
in terms of
what are some of the example use cases and maybe if,
getting into,
you know, where and how does,
how do people access that?
(Micah Callough)So it is in ArcGIS Living Atlas of the World.
You can search that one online.
It's the largest geospatial data
(27:29):
repository in the world.
We spend a lot of money and effort
to make sure that there are cohesive layers,
authoritative information that help the world kind of
be a better place. Ultimately, that's the idea.
And we curate that information
from commercial from public sources,
(27:51):
all kinds of different ways.
And we do a lot of development
for world
wide layers to help the world
think about things like Covid when it spread.
And we were doing a lot of that,
to land use,
to understanding land use across the world.
But then we have a lot of local providers
and even individuals that provide
content through the Living Atlas of the World.
(28:11):
So it's truly,
Jack likes to call systems of systems approach.
The idea is, is with the software,
they can publish the data in an open way,
and everyone can benefit from that information.
And we make that really seamless and easy to do.
We publish a lot of different things
through the Living Atlas of the World.
And one of those things is the deep learning,
(28:34):
toolkit DLTK.
I think if you search for on Living Atlas,
you will see all the models that we put out there.
And those models
range from everything like find the tree canopy
or determine
which trees are here,
or use the segment
anything model to find some tanks in an imagery.
Right?
I can't tell you how many oil and gas tank farms
(28:57):
I had to digitize. At one point in time.
We had to convince them to stop
doing it, in Powerpoint,
I mean, it's just such
inefficiencies in these projects.
And so we were doing it in GIS,
and we thought we were really innovative.
Now with AI, like you talked about earlier, Dave,
that job would be done and done right.
Like I could move from state to state
(29:18):
and get those tank farms,
understand what they're going to be
when you're doing an EIS, when you're starting to
to when one company
wants to buy another,
or when they're responding to an emergency.
We're seeing a lot around building
and footprints and extraction, as a matter of fact,
we've built a layer of buildings
in the Living Atlas of the World,
and it's open source.
It's every building in the whole world.
(29:40):
A little gray box out
shell, which gives you a nice 3D footprint of it.
We used AI to do that ourselves,
so there are just a lot of ways
to pull out information.
There's even a couple in there,
I've seen some recently around, like
looking at if there's a vehicle or a vehicle type
A truck entering a facility or not.
So that
gets beyond the traditional aerial photography
(30:02):
and starts to look at videos.
And we now have the ability to work with video CCTV
feeds, not just imagery.
So you really see this evolving and stepping.
And we put a lot of those models out there
to entice people to recognize that
if they're doing it in a manual way,
here's a better way.
(30:23):
And and I have seen a few projects so far
where a few firms have adopted it
and wiped out the competition pretty quickly.
It is the new way of doing things.
We want to take that complexity
that we were doing before,
and the hours that it took us to do that,
we want to be able to spend those
that time doing something
more creative, more inventive.
We want to move forward a little bit more.
(30:45):
So that's why we stick it out there and we try to, it
helps people to really accelerate what they're doing,
and it's usually their first
taste of AI, if you will.
(Dave Mulholland)Question is, you
know, when they look at that,
when they use it, you know, how do they,
how do they provide feedback to you
and how do you actually get the information
to make the models better?
You know,
because they're there's going to be a lot of users
(31:06):
out there that are going to look at this and go,
wow, there's a lot of exciting tools
that Esri is producing as starting models.
That may want to be a partner
with you to actually evolve.
So how
how do they actually get information
to you or participate to actually,
you know, help these models grow and become smarter?
(Micah Callough)In, oh, so many ways, Dave.
(31:27):
Esri is very, flat, very matrixed organization.
We have things like the, the communities,
the community pages.
There are a lot of communities
that those have direct pipelines
into our product teams.
There's some communities around the
AI and image extraction
and people
that have a lot of discussion
and even help each other out
without Esri even being involved. Right.
(31:49):
There are a lot of channels.
Our product management and product engineering
teams are highly accessible.
If you've worked in or around Esri,
as you've
probably talked
to a product engineer directly more than once,
they're often at our events for communication. Teams
like my team, my solutions engineers.
We are always working through the accounts,
(32:11):
and then we have a much deeper relationship
with companies like VHB at the partnership level,
where we have much deeper conversations
like we did last week around
not only where they think things are at today,
but where things need to evolve tomorrow
to make you successful.
Yeah, I did have some questions around that.
(32:31):
(Scott Jay Ringle)So more specifically, Micah,
I have more on the more technical side
just a little bit.
Anyway,
as far as the models are going, so are there like,
this might be a loaded question, too,
are there maybe 3 or 4 models
that are top, top use cases?
And 70 models is a lot. And are those models?
I'm assuming that some of these are highly specialized.
These smaller models that are tuned
(32:52):
specifically for these exact use cases.
Are those open source?
How locked up
are they?
Can they be trained, fine-tuned
to what levels those kind of things and more
were involved in the actual models themselves.
Are you allowing people to mess with this stuff?
That's what I'm getting at.
(Micah Callough)Very much so.
And we encourage it, right?
That is, the idea is, is that we are
we are putting these out there as a starting point,
(33:12):
not the end point.
And the idea is, is that people will take them
and do a lot of different things.
The big ones
we see now were the obvious ones, people
that we're doing impervious surface
the old school way are now doing it
through extraction.
People that want to extract assets,
like we've seen a lot around the power poles
and the sag and the power lines and things like that.
People that want to build
more realistic scenes with the trees
(33:34):
and stuff like that in 3D,
the Segment Anything Model has definitely been one
because there's a lot around
not knowing what you're hunting for
and just kind of looking for it,
parking spaces and things like that.
But we
absolutely encourage people
to take them, use them and adapt them.
I've seen others that, and even
including myself,
(33:55):
where I've taken them
and adapted them for doing things like extracting,
assets along train corridors, which are,
you're very difficult to get into.
You have to be Rail Safe certified
and using two cameras,
that are GoPros and a GPS point
and the snapping off pictures
every couple of minutes,
which means we can start to extract that data
(34:16):
in a way
that was previously more heavy
and hard to do with LiDAR data.
A lot easier to do with imagery, and we can use LiDAR
data to back up and check our work if we need to.
When we need to up the probability.
So we just see a lot of people
adapting them every day.
I walk into a new customers. Why?
I like to go sit with customers like VHB
(34:38):
because I learned a lot being there and listening
to that, the team and what they're doing,
and there's just some amazing innovations.
It's what I hoped would happen
five years ago is really, truly beginning to happen
is as we focus more on AEC,
they're starting to jump and step faster and faster.
It's amazing.
(Scott Jay Ringle)It is amazing.
(34:58):
And that's really good to hear
that type of collaboration
you're allowing to happen
with these with these models.
Because that's what it takes.
And coming from,
as everyone knows,
like coming from Microsoft as an example,
being able to use open source models
and obviously closed source
models are a big foundational piece of Microsoft,
don't get me wrong.
But the bottom line is,
open source,
smaller models are becoming the way that we see
(35:19):
AI be deployed on a more common,
on a day to day basis.
It's less expensive, it's easier to tune.
You have a better outcome with the result.
It is, the 70 models you have on the allow
the and the ability
to collaborate very closely with the team
and go out and use it and customize it,
I think is a huge step forward
in getting the adoption of AI across,
all the verticals we're talking about.
(35:40):
It's incredibly powerful, actually.
So I'm glad to hear that.
(Micah Callough)Yeah, we're
we're if nothing else,
the Living Atlas, we put a lot of effort into.
The other thing that we do
a lot of
is the open standards groups,
OGC, buildingSMART International.
A lot of the open things,
we don't just say we're open, we put a lot of money
and effort
and we publish open standards for everyone to use.
(36:02):
We're not just about what we do with it.
It's it's literally,
the building scene layer is a good example of that.
I3S, that is an open
standard that anybody in the world can use.
That's excellent.
(Scott Jay Ringle)I love to hear it.
And back to you, Dave.
(Dave Mulholland)So, Micah, I,
wanted to go back to one of the previous comments
(36:23):
we were talking about is really AI,
AEC space, the ethical, conundrum that I think,
that AI could present if not, careful, sensitive,
PII data that's out there.
What conversations should firms be
having around this relative to it
within their organizations?
(36:44):
And what is Esri doing to actually
also helping to actually get that message out?
(Micah Callough)Yeah.
First and foremost, absolutely.
This should be at the center of any conversations.
The security, the privacy information.
I worked for a global firm, so maybe the US
not so far ahead in this is you would see.
(37:06):
We're at a state now where all this is going on.
There's going to be some
there's going to be an injection
at some point in time at the government level.
Right. Like it's inevitable.
It's going to have to be.
It's our responsibilities to understand that.
But AEC firms have an added extra responsibility
because an AEC firm puts a stamp on something
(37:27):
and says,
this is what we're doing, and I'm an engineer
and this is possible to do. Right.
And not all of our work
is about building something and stamping it.
But the reality is a lot of it is,
and or a lot of it comes with the expectation
that an engineer said, I can do this.
Therefore it's something I can do.
So if we're relying on AI that's rolling on bad data
(37:49):
and making decisions
about how I should engineer something
that's not how AI should
or could be used right now, in my opinion.
Now, could it help the engineer
to make some decisions
and think through some iterations of what
they're trying to do
so such that they understand
what that decision was being made from? Absolutely.
So that's my biggest like.
(38:10):
And I get it all the time
because I want to run forward with the technology.
And then the engineers like,
you know, there's
this level of kind of freak out for a moment
and then we all calm down for a minute
and we start talking it through and understanding
what we're doing and what we're using it for.
And we go, well,
you know, this could actually help me.
So you've got to get through that moment of chaos,
(38:31):
I think, in these situations
and then work through, like Scott said, change.
(Dave Mulholland)I again, I did mention
that, you know,
I've been in the business 30 years
and I sit there and look at the parity
between when I started and where it is now.
And there's a lot of fear out there relative to AI.
And I go back to when I started,
traffic engineer by trade,
you know, I ended up having to do a strain
(38:54):
pull calculation for a signal by hand.
I did a calculation by hand,
and then all of a sudden
they handed me a software package that said,
can you do it using this?
It was fractions of the time.
So I relate AI as being very similar to that.
Right.
You're actually having another tool
that you need to actually look at.
Check, validate and make sure that you,
(39:15):
do not get those hallucinogens
or errors that come out of the models.
But realistically, it's incumbent
on us, the engineers in the room,
to actually make sure that happens.
(Micah Callough)I mean,
the other thing, Scott, I'm sure you see this is
some of the reactions
is to stick the head in the sand
and kind of try to ignore it. Absolutely.
(Scott Jay Ringle)That was one of the blockers.
(39:36):
We were talking about that just yesterday, in fact,
not taking action or not paying attention
or thinking this is another fad
or just not wanting to adjust and change.
Absolutely. See it a lot.
That and data.
That's one
and two blockers
right there for implementation of AI.
(Dave Mulholland)So I was going to take the step forward now.
(39:56):
We mentioned there's three different areas.
I want to start to look forward into,
the next five years,
in terms of some of the significant advancement.
But so, Micah, I'm coming to you Esri, GeoAI, Launch,
where are we going to go over the next five years?
And where do you see it going?
And what do you hope to,
you know,
get accomplished over
the next five years in the AI space?
(40:18):
(Micah Callough)Yeah.
So I'll, I talked a lot about GeoAI because that's
the immediate low hanging fruit,
the thing that people can wrap
their hands around right now.
There are also smaller things that are happening.
And I can't say that they're smaller.
They're just more behind the
scenes and more ingrained.
One, for example,
is we were building a lot of AI into our systems.
(40:39):
I talked about that idea
of weaving stuff in AI into the product.
Survey123 on the,
I can't remember what version of,
is the latest beta at this stage.
Or it might be the latest
production release, has an AI function
that I used a couple days ago.
I asked it to create me a form for a webinar
or a meeting
I was going to have, where I wanted feedback on
(41:00):
blah blah
and I fed it some information
and boom, it created me a form
that I could send out to do a survey on.
So that would have taken me
an hour or two to think and come up with.
Did it, was it perfect? No.
There's a couple things I threw out of there
and some stuff I added
because of my knowledge
and what I'm seeking
as far as information
that I didn't prompt
the AI for, I didn't feed it enough good information.
(41:23):
But man, it did get me pretty far pretty fast.
And we're really thinking
about that across all of our products.
Back at the Federal User Conference
earlier this year,
some of you may have seen a, an AI prompt tool
that was designed to allow people to ask questions
that are geospatial
and not have to go to a GIS person to get the answer.
(41:43):
So, you know,
I want to understand my property,
you know, what's the value of it?
Where you know what, what's within walking
distance of this area?
Is this property commercially viable, right?
Just being able to do things
and get to geospatial information
without knowing
what geospatial is in the first place.
So those are a lot of the assistant-type models.
(42:05):
And then underneath the hood,
we've talked a lot about people
who build applications.
And we have to know to turn this layer on
and this thing and click this tool
and click this button.
An engineer would probably just rather
ask it a question and get an answer, right.
They don't want to know about those buttons and tools
and all the things that are going on.
So how do we make take the AI
and the assistant type chat models
(42:27):
and make that better and faster?
So that's a little bit
of where Esri is focusing its attention.
There's a lot of deeper R&D things
that are, that are getting really interesting.
And we're just kind of
in that balancing game of like, where are the models?
Is GPT the thing and always going to be the thing?
I'm not so sure it will be, right, long term.
(42:47):
Scott can help me with that.
(Scott Jay Ringle)I think AI will completely transform
how we design and build
by enabling real time
dynamic project design and execution.
And that sounds like a marketing term,
but I want to, when I mean that,
I mean that sincerely.
I see a future
where AI is integrated into every phase of a project,
from concept to construction.
(43:08):
It's basically
AI-driven systems will continuously analyze data,
adjust designs,
and even make real-time decisions on site.
So imagine this, if you will,
a condition changes on a construction site,
a construction site, whether it's weather,
supply chain disruption, unforeseen challenges,
(43:30):
AI algorithms will automatically adapt project
plans, reroute resources and adjust timelines.
It will continue with the overall goal
of the project timeline
and keep all this in mind
at the same time, simultaneously, which is a,
this will eliminate
many of the delays and inefficiencies
that plague the industry.
A huge piece of where this eventually is
(43:50):
headed to where there's
so much automation, there's people involved.
It's a copilot,
if you will, not to overuse that word,
but it's a copilot for the people
that are involved in these projects
at all aspects and everyone
co-developing this together.
But the AI streamlines the process to that level.
Architects, engineers,
contractors will all be alongside
with your AI assistant, allowing the the creativity
(44:13):
and the efficiency to be enhanced ten fold,
100 fold.
In short,
if I was to summarize this,
the future will be a blend of human expertise
and any AI-driven precision
to create a more responsive process for design
across the board and architecture.
I mean, that kind of philosophy
is where I'm going with this, if that makes sense.
(Micah Callough)I 100% agree.
(44:33):
I saw that,
I was at a construction event recently, and I
and they had an
AI as one of the topics during the afternoon
they were talking about.
And I kind of thought, well,
this will be interesting, right?
Like these guys aren't using it
and they were quite well using it for like
figuring out billing issues, looking at risks stuff.
They were very much adapting about what
they were doing,
which was really good, was choking it
(44:55):
down to the problem that they have.
So constraining it for risk.
But I find it interesting that it's, it is it's the
this is the fastest I've ever seen
anything get adopted in this industry right?
Like ever.
And more people are excited about it
that I, that I think barely can open up
Excel or Word. Right.
(Scott Jay Ringle)That's so true.
(45:16):
(Dave Mulholland)I do want to pivot and actually go
with a closing question.
And this is a, this is a general one.
One takeaway, right.
If you had to advise a firm that's just now
getting into AI, that's just now starting to use AI,
what is that one action that they should take today
(45:36):
to actually prepare
and get going into the AI space specifically?
And you know, Micah,
it can be using it, the GeoAI
specifically if you like.
(Micah Callough)Well,
you know,
as much as I'd love
to sell GeoAI and ArcGIS, I'm actually going to say
it's a little bit more about
(45:57):
bringing, it's the people, right?
It's going to be about the people, right?
I get it's a lot of people management,
a lot of change management.
Getting the right partners,
getting the right staff brought in
that has the experience that can help
accelerate the rest of your staff,
helping the people
to understand where it's going to be.
Technology is usually never the problem.
(46:18):
It's usually the people transformation.
And I know that from experience,
I have been beat up
more than my fair share
trying to implement technology inside of an AEC firm.
But I think now is the right time
and there are people out there that are really,
they love to solve these problems, right?
So work with them,
figure out
how you partner with them,
(46:39):
and by all means leverage GeoAI if you want to try
to start doing something, right, like that's
a good starting point for some of those folks.
(Dave Mulholland)So, so as
as the engineer in the room, I agree with you.
I think GeoAI is actually one of the first entry ways
and easy
opportunities for a firm
to actually start to engage and embrace.
(47:00):
And I'm looking forward to, to the opportunity
to actually
use it on
some of the activities
that we do, across the entire footprint.
Scott, the one takeaway,
what's that one action
that you have to basically advise the listeners to,
to, to take note of.
(Scott Jay Ringle)Don't be afraid to move forward.
And it comes down to a single small little word:
(47:22):
data.
Talk to us about data.
Don't be afraid to have your data prepared and ready
to move forward.
And again, that's a general statement.
But the the net net of this is
the single biggest blocker, and to Micah's point is
that people,
the people are already in the mindset
of being ready to move forward with this technology.
(47:44):
The number two blocker is going to be
their data assets and their data state.
So we need to talk about that
and demystify that complex data that they have,
which is not as hard as it sounds.
There's a lot of fear,
uncertainty and doubt,
as they say in sales terminology,
wrapped around data assets.
It just takes someone that knows how to analyze that,
come in, have a conversation, do a discovery workshop
(48:06):
to get you ready
and prepared to go down that journey to build out AI
applications and platforms at scale,
because it all starts there.
From there, it becomes magical.
That's no pun intended,
but that's that's my takeaway, Dave.
It's going to be analyze
data and data data data it all starts there.
(Dave Mulholland)You cannot do AI without good data.
(48:28):
And I think, Micah, you referred to... Good data.
You refer to that.
(Scott Jay Ringle)Yeah. Good data. I apologize.
Yeah. You know good, good data there.
I was sitting there and I was reflecting
and it's one of those general takeaways.
I think it's the eyes wide open, right?
Relative to where AI is in the industry is,
I was going to come back with a, you know,
my response was the policies
and be aware of what's going on within it.
(48:48):
Micah,
you referred to the state agencies
as some of the regulations that are coming.
I think for folks
that, would be listening
to this, one of those key messages
that we got out is
agencies are already starting
to look at this, already
starting to understand the impacts of it.
When you think about procurement
and going after a project, oftentimes
now they're including some provisions
(49:08):
relative to the use of AI in your pursuit.
Be careful.
And some of these are saying you can be disqualified
if you use AI as part of your submission.
Same thing on the project delivery.
When you think about like scoping a project
and scoping
and delivering a project for a client,
some of the state agencies across
the US are now looking at if you do not inform
(49:31):
the PM that you're using AI as a delivery,
then you could potentially be
disqualified from actually this submission.
So it's a it's an eyes wide open for me.
It is here.
It is moving.
I think both of you indicate
it is moving exponentially.
Micah, I'll be honest with you, I'm
looking forward to seeing the excitement of where
Esri is taking this over the next 5 or 10 years,
(49:53):
when I actually get my next 30 years in
in the industry.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.