Photo: Pixabay
“As the cost of deploying large-scale camera networks continues to fall, and the powers of artificial intelligence continues to rise, we’re going to see these types of camera systems be taken out of the hands of the state, and go into the hands of corporations,” says Andrew Chen.
Photo: Wikimedia Commons / Paweł Zdziarski
Why? To gain insights into how their consumers are using physical spaces.
“For example,” he says, “Let’s say you own a supermarket. And there are a bunch of things about how you set up that supermarket that are shown to have really strong effects on consumer purchasing behaviour. Things like how you arrange your shelves, how easy it is for people to get in and out of your supermarket, right?”
Up until recently, most of these insights have come from stationing human researchers in the supermarket and with a clipboard and a pen making notes about people as they shop. The downside of that approach is that not everyone can be observed all the time, and when people know they’re being observed, they tend to behave differently.
A camera network, linked to powerful face and body-recognition software, fixes both of these research problems.
Photo: 123RF
The consequences?
“At a very low level, you might be able to do things like figure out how many customers you have in your shop. Which paths they are taking. There are commercially-available systems right now that can analyse checkout queues. And tell the manager that the queue is getting too long and that they need to open up a new checkout counter. So that’s probably a good thing.”
However, the degree of granularity of the observations is greatly increased by technology. Business owners can start to collect statistics over time, and you can answer high-level questions like “which products should I put closer to the entrances and exits?” How often do we need to restock certain aisles? How many staff do I need on a certain day?”
Photo: 123RF
If those topics might seem reasonable and not unduly invasive, how about even higher-level questions like “Which items did loyal customer 362 pick out today but ultimately not buy, so that we can send them an email with a special offer for next time?”
If we are uncomfortable with that level of scrutiny, how about the question “Is this person who has just entered at risk of shoplifting based on their past criminal history?”
Or “If people who look a particular way tend to buy more stuff, should we send a staff member to upsell to them?”
A woman shopping. Photo: 123RF
Then there this the rich, and richly profitable area of secondary uses of this information. What happens if the supermarket sells some of that data to a food manufacturer? Or to a health insurance company? Or provides direct links to that footage to the police?
Chen thinks that the reason these scenarios seem scary is that we like to pretend that technology is value-neutral. That it is just a piece of software or hardware. A tool, not inherently good or bad.
“But in some senses,” he concludes, “this isn’t true.”
He thinks that a key factor here is that the owner of the surveillance system is a commercial owner, rather than an agent of the state. When it comes to corporations its incentive is not to serve the public good (which it can be argued is the motivating factor for government deployment even if its use is controversial).
“Companies aren’t using the camera networks for your safety. The benefit isn’t accruing to you. They’re using it to find ways to make them more money.”
Andrew Chen Photo: University of Auckland
About the speaker
Andrew Chen is a PhD Candidate in Electrical and Computer Engineering at the University of Auckland. His research focuses on embedded vision, helping computers, namely robots and surveillance systems, ‘see and understand’ the world around them in real-time. In light of this, Andrew has a keen interest in surveillance and privacy implications, particularly how we can use technology to better protect privacy and avoid abuse by system owners.
Andrew also teaches digital systems and software design, and sits on investment committees for ReturnOnScience to identify and commercialise innovative, entrepreneurial ideas.
This session was recorded at the University of Auckland’s Raising the Bar night in August 2018
Photo: University of Auckland
Raising the Bar was recorded in association with the University of Auckland
