
Italy’s social insurance relies on Articfial Intelligence to combat fraud
Human beings to take last decision.
Dr. Sch.-W. – 11/2019
It’ll
make everybody’s heart beat faster when accessories from the icy world of
“Frozen” are spread under the Christmas tree. This will be less true of
economic actors who operate across borders if the bloodless algorithms of the
software “Frozen” prove they have violated EU Posted Worker Rules. On the basis
of “big data” analyses of over a million A1 forms, the software creates a
predictive model which will be able to identify the greatest risks of illegal
activities and to reveal such instances.
The
executive director in charge of “Income and Contribution, Risk Analysis and
Fraud Prevention”, Antonella Lilla of the Directorate General of the Italian
social security institute INPS, expounded interesting details, especially with
reference to the transition from repetitive functions to “self-learning
intelligence”. In order for the software to fulfil its functions, continuous
training based on current cases is necessary, i.e. cases where collaborators
have revealed instances of fraud and have drawn conclusions. In view of the
close cooperation between man and machine, Antonella Lilla describes this tool
as being a “system of collective artificial intelligence” that keeps on
learning in a circle all the time. It submits concrete suggestions to collaborators
in charge for their decision-making. In all these cases, it’s Man who has the
final say. Suspicious cases lead to further inspections on the basis of which a
final decision will be taken, if appropriate. In return, the latter will be fed
into the “Frozen” system, which does not appear to be that “frozen” at all.
Maybe
some other information may be inferred from this model that can be applied to
other use cases in the field of social insurance. The enormous amount of
individual-related social and other data has been a source of concern. Handling
these risks will be one of the big challenges when dealing with ‘big data’.