ST poll: Safety before privacy in AI surveillance

07 Sep 2019

Straits Times, 7 Sep 2019, ST poll: Safety before privacy in AI surveillance
Privacy takes a back seat in Singapore when it comes to safety and security, a Straits Times poll on attitudes towards the use of artificial intelligence (AI) shows.
But the technique that allows machines to learn from enormous sets of data is less acceptable when used to determine people's credit worthiness or work productivity.
More than two-thirds of some 5,000 respondents in last month's survey supported the use of AI in law enforcement and drowning detection.
Specifically, respondents were asked if they supported the use of scanners that recorded and matched people's faces and movements against a citizen database to help nab law breakers.
They were also asked if they agreed to the use of video analytic tools to detect the onset of drowning so rescuers could be alerted.
Close to two-thirds of respondents, however, objected to the use of AI tools to determine workers' productivity, say, by screening keystrokes on computers and the duration of breaks.
Similarly, close to two-thirds of those polled would not let AI technology comb through their entire digital footprint - including social media use, Internet browsing habits, location data and other smartphone information - to determine their ability to pay back a loan.
The questions were part of an interactive quiz launched last month to assess how ready people were to yield control of their lives to AI, which has moved into homes, streets and workplaces.
Singapore University of Technology and Design (SUTD) communication and technology professor Lim Sun Sun said she was not surprised that people here viewed AI surveillance for safety and security positively.
Prof Lim, who also heads SUTD's humanities, arts and social sciences department, warned against pre-emptive policing though.
"If AI surveillance becomes too intrusive to the extent that our freedom to be left alone is encroached upon, such as with pre-emptive policing, people will have reservations," she said.
Other types of policing, such as in determining employees' productivity or one's credit-worthiness, were frowned upon.
Mr Hagen Rooke, counsel at London-based law firm Reed Smith, said that these AI systems do not capture all the data required to appraise an employee's performance.
"Similarly, scoring one's credit worthiness based on one's digital footprint may unfairly distort the credit profile and act as a socially stigmatising factor," said Mr Rooke.
Furthermore, the lack of transparency around such credit scoring algorithms compounds the sense of insecurity towards the technology, said Prof Lim.
At Thursday's Sooner Than You Think AI conference at Conrad Centennial Singapore organised by Bloomberg, Minister for Communications and Information S. Iswaran said that trust underpinned all AI efforts.
"Ultimately citizens and individuals must feel that all these initiatives at the end of the day are focused on delivering welfare benefits for them," he said.