Lawrence Weber, Managing Partner for Innovation at Karmarama, on facial recognition technology and the three questions brands should consider before implementing new tech.
At the start of every Karmarama client innovation workshop we play a game called ‘creepy versus cool’. The object of the game is to judge the use of new technologies in the abstract and then again with an example use case in mind.
The exercise is a light-hearted way to warm up people’s minds and vocal chords. But it also has a more serious intent. It’s a useful way of reminding everyone that the new products and services they dream up shouldn’t be judged solely on what’s feasible and potentially profitable, but also on their potential ethical and societal impact too.
We believe that technology can be a positive force and help to deliver good works (which is Karmarama’s own ethos too). But we also firmly believe that just because we could use a new piece of tech, doesn’t mean we should.
So to warm you up, I’d like you to play along too.
Firstly, decide whether this statement is creepy or cool: “your national government is using publicly available social data to encourage more responsible behaviour.”
We normally get a 70 to 30 per cent vote in favour of cool in response to this question.
Once you’ve made your decision, have a think about this particular use case: “China is using facial recognition technology to send jaywalkers fines through text messages.”
At this point the votes swings to 80 per cent in favour of creepy.
When we ask people why they voted creepy, they say that despite the end outcome feeling positive, i.e. less people getting hit by cars, the method used worries them deeply. When probed, it’s specifically the use of deeply personal data to power artificial intelligence (AI) that is used to identify and make judgements about people – all without any human oversight – that really concerns them.
As more technologies are developed that can gather data from our physical attributes and actions, that concern will only grow. Gartner picked up on that trend in January 2018 by predicting that by 2022, personal devices will know more about an individual's emotional state than his or her own family.
So what does this mean for brands? On the one hand it feels like understanding consumers better might well lead to more relevant targeting, better products and more empathetic customer service. On the other, the ability to be creepy without meaning to has never been greater.
Here are three questions we think brands should ask themselves about every new idea they are considering:
Is there a clear benefit to our consumers, and do they understand it?
While it might be possible to use facial recognition to identify a repeat visit to a retail store and reward a customer with personalised offers, it’s possible that it could create a negative reaction. That’s partly due to the nature of the technology involved, but it’s also linked to the value sitting too squarely with the brand. However, using machine vision technology and a patient’s medical history to identify undiagnosed skin conditions is a service many people would actively pay for. The context and nature of the benefit is key.
- How well-trained is our AI?
The potential benefits of brands giving some decision-making over to an AI are clear. And those benefits can certainly be passed onto consumers. However, the examples of machine vision AIs not recognising famous black presidents and sportsman, or stopping non-white employees entering meeting rooms, highlight the massive issue that this kind of automation can create. Being clear about what data you are going to train an AI with and where your data is lacking is crucial, and should highlight some other things that need fixing in your company too.
- When do the humans get involved?
It’s a valid question outside of the creepy versus cool, new technology debate. But it’s certainly one that should be top of mind when designing a system that takes data from humans and uses a logic system to process them. It’s very easy for any one of us to misinterpret the actions of another human being and therefore very easy for us to unwittingly build in those failings into an automated system. Building in and celebrating the human element in an automated system is another key attribute to avoiding unexpected consequences.
Taking advantage of new technologies like facial recognition feels like something no progressive brand will want to pass over. And nor should they. But a good empathetic brand will always realise that being cool is better than being creepy.
Try that simple test on your next idea and see if you pass.
Lawrence Weber is Managing Partner for Innovation at Karmarama, part of Accenture Interactive, chair of the IPA BrandTech group and a director of Innovation Social.
Last updated 24/05/2018