The Queensland Police Service (QPS) plan to trial artificial intelligence (AI) systems to ‘predict’ domestic violence before it occurs.
The trial, set to begin in some police districts before the end of 2021, proposes the use of predictive analytical tools to identify escalating situations and intervene before a ‘point of crisis’.
Under development for about three years, the tools aim to predict behaviour based on previous calls to an address, past criminal activity, and other police-held data. By preempting escalations behind closed doors, acting Supt. Ben Martain states that authorities will be “proactively knocking on doors without any call for service.”
If this all sounds a little like Minority Report and you’re wondering when Tom Cruise is going to show up, you’re not alone.
Concerns of misuse and unintentional impacts on domestic violence victims have prompted Martain to assert that the system could not be used for Minority Report-style arrests, or as evidence in court.
And while the tools may not be as farfetched as the use of clairvoyant humans in the 2002 film, the impact on the victims of domestic violence has the potential to be very real.
Some initiatives are explicitly not Minority Report or The Circle styles, yet clearly steps toward them. Let’s brace ourselves for tensions between needs and challenges. But frankly, I bet better education, respect of the Other and Love can do better. https://t.co/SvvWooX0LW https://t.co/x44qQ5FYCe
— Eric Platon (@not_replica) September 14, 2021
One major concern is programs that extend surveillance of a perpetrator will also deepen the surveillance of victims.
The autonomy of victims is also at stake, as police intervention is not always welcome. Victims becoming misidentified as perpetrators is sadly not an uncommon occurrence and victims who have used violence in instances of self-defense have found themselves arrested instead of the perpetrator.
There is also the risk that uninvited door knocking can itself lead to an escalation as visited persons may use offensive language or refuse to identify themselves, which can, in turn, lead to charges.
Outlined in a Human Rights Commission report into Indigenous Deaths in Custody, the ‘trifecta’ phenomenon can occur where the subject of a door-knock, who has not otherwise offended, may be charged with offensive language, resisting arrest, and assaulting police.
Professor Lyria Bennett Moses, the director of the Allens Hub for Technology, Law and Innovation at the University of NSW, says a similar ‘knock on doors’ approach in NSW resulted in Indigenous youths being targeted.
I might add that the reference in the quote from police to objective data belies the promise to look for bias. https://t.co/XXgItyVvff
— Lyria Bennett Moses (@lyria1) September 14, 2021
Advocates for domestic violence victims agree that there need to be better systems in place to identify and prevent high-risk offenders.
Campaigner Angela Lynch says of particular concern is “those who go from relationship to relationship.”
Martain has assured the public that the “QPS considered the lessons learnt in other jurisdictions and have developed a model monitoring tool that aims to regularly monitor and address bias within the model.”
He stated, “For the pilot, QPS removed raw data that had the direct attributes of ethnicity and geographic location before training the model.”
Bias mitigation will be the focus of a research project that will operate alongside the AI models before they are rolled out.