Page 16 - Are You Future Ready?

It is also difficult to interpret analytically. As with the
pixilation problem in remotes sensing technology,

the finer the detail, the harder it can sometimes be to
interpret. When the model is a large number of interacting
multi-dimensional decision agents, sensors or social-
media data points, what-if simulations are governed by

many parameters, which make results unstable, highly
dependent on starting conditions, interactions between
parameters and so on. The attractiveness of the 1:1
model’s complexity becomes a liability when it comes to

014 modelling cause-effect and interpreting results. Historically,
scientists and philosophers have abstracted for a good
reason. It is difficult to understand the way the human
mind thinks, for example, without a robust abstraction of

the idea of morality or learning or human judgement. The
multiplicity of ‘raw’ information in a digital twin model of
a city leads us, ironically, backwards (or is it forwards?)
towards black-box thinking, where we observe systemic

behaviour without being sure of the contributions of
individual variables and parameters.

Another sense in which more complexity leads to more

black-box thinking is in the application of AI and advanced