05 | GEOSPATIAL STRATEGY ESSENTIALS FOR MANAGERS MATTHEW LEWIN
The machine learning model is fed a series of
datasets containing images of damaged and
undamaged properties. The AI then develops
a pattern recognition model comprised of
variables of its choosing. Based on the model,
the AI predicts which properties appear to be
damaged and which don't. For a human to do
this, it would be very time consuming. AI makes
this process a snap and the resulting analysis is
usually much more accurate.
When you put them together—more-readily-
available spatial data and cheaper, faster, AI-
driven spatial analysis—you get better spatial
awareness at a better price.
Implications for People and
Organizations
What does cheaper spatial awareness mean for
people and organizations? When something
valuable gets cheaper, economics tells us
two things happen: 1) people start using it
in non-traditional ways, and 2) other factors
change in value depending on whether
they're complements or substitutes. (Watch
this excellent presentation to understand this
concept in more detail.)
The Expanded Role of Spatial
Awareness
As new machine learning techniques and
ubiquitous spatial data lower the barriers to
spatial awareness, we'll start to see a deeper
application of geography to long-standing
problems. Consider human movement.
Understanding why people travel from one
location to another is notoriously difficult. Much
of this understanding depends on scale—are we
interested in short-term, localized movements
or longer-term, regional migration patterns? In
both cases, the motivating factors are varied and
complex. That's why predicting where people
will travel to is so hard.
Recently, we've seen advances in both cases.
On the local scale, researchers have developed
spatial AI models to monitor street cameras
and adjust traffic lights to help disperse crowds
after live public events. Traditional methods of
planning exit routes are burdened by trying to
account for all the possible reasons people go
one direction or another, and on-the-ground
crowd control tactics can be inefficient and
quickly overwhelmed. The new approach
analyzes real-time video to detect areas where
crowds are forming and predicts which routes
would best ease congestion. Controllers can
then adjust traffic lights accordingly. The AI
provides a much more dynamic level of spatial
awareness, and actively enables public safety.
On the global scale, researchers have—for
the first time—developed models that predict
international migration patterns resulting from
climate poverty. The models incorporated
environmental, socioeconomic and political
factors into different climate scenarios that could
affect regions of Central America. Included in the
models were factors like carbon concentrations,
GDP and border management. Over 10 billion
data points were fed to a supercomputer,
which developed predictions of how different
populations would respond to different climate
change scenarios. The models were compared
against historical migration patterns to assess
and refine their accuracy.
The result was five migration projections
that illustrate the profound impact of climate
change on population migration—specifically
how it could drive the displacement of huge
populations in the next 50 years. In the
past, projections like these were impossible
considering the cost of computing and data
acquisition, and the relative immaturity of
spatial modelling algorithms. In short, the price
of spatial awareness was just too high. In the
future, we will almost certainly see more of these
breakthroughs.
