Police in the UK will pilot a test run of a new AI system which intends to track hateful comments on social media and identify the physical locations of those posting the tweets in an effort to predict and prevent hate crimes.
The AI system was developed by researchers at Cardiff University as part of the University’s HateLab project. The project collected information from Twitter and from local crime statistics in London between August 2013 and August 2014 and found that as “hate tweets” rose in frequency in a given area, hate crimes in the area would subsequently follow a similar increase. Hate tweets are classified as being “antagonistic in terms of race, ethnicity or religion.”
Now, the tool is being taken out of the lab and will be deployed for public testing on October 31st, the date that the UK is set to complete ‘Brexit’ and leave the European Union. Professor Matthew Williams of Cardiff University stated that this event was intentionally chosen, as the discourse surrounding the event will allow the AI to operate in a “toxic” political environment:
“Brexit is one of our test cases to see if hate speech will spike. There has been talk of riots on the streets, and there is an expectation that tensions will bubble up around that date.”
A National Police Chief Council (NPCC) spokesperson claims that the AI will not be used “to ‘trawl’ for offences” but instead aims to “intervene at the earliest possible stage to reassure victims that we will act to protect them.”
“It is not used to ‘trawl’ for offences but will help us to identify early rises in tensions, informing force level operational decisions and our work with communities in seeking to reduce hostility at an early stage.”
“Hateful abuse online can leave victims with significant psychological harm, but can also lead to more serious physical offences. Police need to be able to intervene at the earliest possible stage to reassure victims that we will act to protect them.”
“We absolutely recognise and uphold the right to free speech even where it causes offence – but this does not extend to inciting hatred or threatening people.”
The AI will be included as part of the polices’ ‘hate speech dashboard,’ a tool used to allegedly assess ‘potential threats’ to communities. The dashboard received “£1.7m of funding from the Economic and Social Research Council (ESRC) as well as the US Department of Justice” according to The Telegraph.