In our last
article, we wrote about the concept of precrime, its definition and origins,
and we related it to technology ethics. In this post, we are going to dig deeper with two
real-life examples of how precrime mechanisms are used to detect criminal law
violations around the world.
Sweetie, the virtual bait for paedophiles
In 2013, in the Netherlands, the
children’s charity Terre des Hommes carried out a precrime experiment
against paedophilia with surprisingly good results. They used a
computer-generated fake profile to pose on video chat rooms as Sweetie, a
10-year-old Filipino girl. They did this during 10 weeks, in which around
20,000 men contacted the profile, 1,000 of them trying to pay her money to
perform sex acts online. Sweetie was so life-like that online predators were
caught within seconds after the charity members logged in the chat rooms.
The
researchers used evidence as profiles on social media to identify the suspects,
and these findings were then passed to police. The charity also wanted to
provide authorities with the developed technology.
However,
different views were expressed after the experiment. Project director Hans Guyt
declared that this type of crime "requires a new way of policing". On
the other hand, police authorities themselves, such as Europol and UK’s
National Crime Agency, said that such an investigation should be best left to
specialized law enforcement agencies. In this example we can see a difference
in ethical opinions within the institutions which resulted in Sweetie’s
deactivation (despite her good numbers).
Veripol: in search for false police reports
In Spain, an
international group of researchers has developed Veripol, a tool for detecting
false police reports based on natural language processing and automatic
learning. One of the experts involved is Miguel Camacho-Collados, a Spanish mathematician and former police inspector who has participated in several
precrime projects in both his country and around the world.
The target
reports for the study were those concerning violent and intimidating thefts, as
the fake reports for these offences seem to have increased during the last
years.
Veripol first
learns a statistical model by comparing two sets of previous reports, true and
false. After this automatic learning, it carries out a linguistic analysis of
the claimant’s statements to determine its verity, based on different parameters.
For example, it is known that fake reports usually insist on the stolen object
and leave out details about the suspect and the situation in which the theft
occurred.
The trials
started in June 2017 in two Spanish provinces, Murcia and Malaga, where the
agents were trained to work with the programme. In only one week, 80 false
report cases were detected and resolved (31 and 49 respectively), which in
comparison to the numbers of the previous eight years in those areas resulted
in an efficiency of 83%. This success has led to a gradual implementation at a
national level that is actually taking place.
Of course, as
the researchers pointed out, fake reports bring several problems to law
enforcement efficiency, and are an offence themselves. But from an ethical
point of view, it could also be alleged that there exists the future risk that
other non-linguistic legal factors, such as witnessing and evidence, could be
taken less into consideration in the investigation process.
As you can
see, precrime is not a mere science-fiction hoax, and real efforts are being
invest to develop advanced crime prediction technologies. However, technology
ethics still have a lot to say when confronting scientific results to real
applications in this field. If you want to know more about the crossroads of
law and technology, follow us… and behave well!
References:
AGENCIA SINC.
2018. Inteligencia artificial a la caza
de denuncias falsas. [Retrieval: 14th January 2019]. Available
at: https://www.agenciasinc.es/Noticias/Inteligencia-artificial-a-la-caza-de-denuncias-falsas
CRAWFORD, A.
2013. Computer-generated 'Sweetie'
catches online predators. [Retrieval: 14th January 2019]. Available
at: https://www.bbc.com/news/uk-24818769
MARTÍN, D.
2017. El español que quiere implantar la
Policía Precrimen de 'Minority Report'. [Retrieval: 14th January
2019]. Available at: https://www.elmundo.es/tecnologia/2017/07/21/5971a077e5fdea7a618b45bc.html
QUIJANO-SÁNCHEZ,
L.; LIBERATORE, F.; CAMACHO-COLLADOS, J.; CAMACHO-COLLADOS, M. 2018. “Applying
automatic text-based detection of deceptive language to police reports:
Extracting behavioral patterns from a multi-step classification model to understand
how we lie to the police”, en Knowledge-Based Systems, Vol. 149. Available at: 10.1016/j.knosys.2018.03.010