towards the end of the month, therefore they fare-
evade.
What is the difference from a student carrying out
the above investigation as part of the Digital Lab to
the conventional desk research? First of all it is the
hands-on approach: the student is not anymore
building theories in his head but tries to bring them
close to the reality. Secondly there is certainly a
design thinking approach in terms of building a wider
context that involves both gnoseological as well as
epistemological aspects. In the example above, he is
also cohabitating the Digital Lab with his girlfriend
that brings knowledge from her own experiences. The
idea is that students get used in building algorithms,
demystifying their power and building the path to
explain decision making as a sequence of human-
made syllogisms. Improvement or partiality can be
embedded within them, same as humans suffer from
bias-variance trade-off (Gigerenzer, 2009).
There is no objection that we need to prepare well-
educated and skilled professionals for the new digital
eras ahead of us, that will be capable to cope with all
the challenges related to the development of new
algorithms and approaches that improve the accuracy
of Artificial Intelligence and Machine Learning
applications, and the creation of software services and
Apps that will have the capacity to tame the triptych
of density – immensity – complexity that appears in
most aspects of today's business fields. Below we
describe for each of these aspects the type of
innovations and research contributions that the
Digital Lab introduces and validate as part of its
operation:
Density: While as practice shows convolutional
neural networks are designed for dense data,
there is a plethora of data that are often sparse
and which cause difficulties in their conversion
to a denser form. In the Digital Lab, we take
advantage of the computing capabilities offered
by edge AI to not interfere in the original pools
of the information, as a great part of the problems
caused by faulty operations are result of
inappropriate densification of the original data
sources. Same as in the literal ecosystems, data
lakes apart from the efficiencies they bring, also
create problems and cascading failures involving
also ethics and privacy aspects, that are difficult
to cope with in later phases.
Immensity: Immensity of data considered as part
of typical Big Data analytics ‘ processing
routines’ as these appear in many real-world
applications in the areas of logistics,
telecommunications or smart cities applications
is not a problem at all for the computing
hardware. However, what we consider as a
comparative advantage with respect to other
approaches is the inherent support that the
Digital Lab philosophy offers for the support of
transparency - explainability - auditability of all
processing routines and algorithms conducted by
the students as users of the Digital Lab.
Complexity: Use of computing hardware may not
reduce the complexity of all processing routines
and algorithms conducted by the students as
users of the digital Lab but shall help them better
understand the problems and issues at stake,
make sense out of them and also help, where
possible, tame the underlying complexity. More
specifically, for the aforementioned case of the
free-riders problem in the Bern transport
network, it might offer an incentive for the city
not only to increase the controls but to better
understand or earlier identify a social problem
that might help avoid future controversies or a
potential growing phenomenon related to social
segregation.
In the above context the Digital Lab may act as a
nursery for new ideas in a variety of fields other than
the economy and the technology, offering the means
for hands-on applied sociology and experimental and
behavioral economics.
We are all of us aware that AI Technologies are
data intensive, so in this respect data literacy is a
must. To this, the acquisition of some basic skills by
the students to understand the rules and the legal
limits in sharing and trading data, have some
familiarization regarding how data privacy issues
force, benefit or prevent the partnership between
companies, or how existing or newly appearing
business constellations can create value and trade
data-based solutions shall better prepare them in their
first steps of their professional career paths.
4 DISCUSSION
An aftermath we come up with regarding our
experiences and experimentation so far with the idea
of a Digital Lab as a core component for our teaching
and research activities at the Business School of the
Berner Fachhochschule is that one should not buy (or
respectively: sell) visions but invest on substance.
One should not care if what may have started as an
applied Artificial Intelligence project may have ended
up as a Big Data analytics endeavor. Same also if
one’s Big Data analytics project ends up in something
related to … Small Data. The important aspect is if it
makes sense.