Please wait while the transcript is being prepared...
0:00
Hi. My name is Amalia Barthel.
I'm an advisor, consultant,
and educator in
the areas of digital
risk, digital data risks,
privacy compliance,
and governance.
0:13
In this talk in the
digital risk series titled
digital risk Implications in
AI and machine learning models,
we will explore the
implications: good and
bad, and the risks of AI
(artificial intelligence) and
such models and also delve a bit
further into who takes
ownership and bears
the responsibility
in organisations
and identify some great
resources for professionals.
Beware of the bias.
Let's now review
this case study on
the use of data from
the internet from
the European Union Agency for
Fundamental Rights also known
as FRA and their paper.
Their quality and artificial
intelligence mitigating bias
and errors to protect
fundamental rights.
The internet is one
important though
not the only source for
data generation and
collection that AI draws on.
Data can come from a
variety of sources
cover different types of
data and various topics.
Social media platforms are
another source for
data generation, but
not the subject of this
particular use case from
the EU (FRA) Fundamental
Rights Agency.
Data from different sources
can impact on bias.
This section focuses only on
data from the internet as
a generic case
study to illustrate
the potential for errors
of representation.
Many businesses try to
harness information
from the internet.
As such, data are often
freely available.
There is not much
information available about
which data are exactly used
for which applications.
However, the Internet and
social media is one of
the frequently used sources.
The following describes
comparative data on the
use of internet data by
companies and
highlights the bias in
internet data at a
general level in the EU.
This should give a
general sense of
coverage issues of data
from the internet and
its potential bias.
While increasing
coverage, data from
the internet may only
reflect the subset of
the entire population, which
is related to limited access to
the internet and
different levels of
participation in online
services such as social media.
