“There is something profound about being able to change a life.” – Dr Vivienne Ming
Last week I wrote about The Promise of AI, sparking a lot of debate with friends and colleagues about the future of our world and what we must do to shape it as artificial intelligence becomes more pervasive. Many were adamant Government and leaders around the world should step in and stop innovation before it goes too far – that there are some AI-enabling technologies that should never see the light of day for fear of what they could become in the future. But at what cost? The ubiquitous end to AI-driven innovation shouldn’t be a Yes or No question, but a look at the broader ethical implications of such decisions and the opportunity cost of making the world a better place.
One of the best examples of this given by Dr Vivienne Ming, Founder and Chair at Socos Labs, is the use of Google Glass and AI-driven facial recognition. For many years Google has wanted to keep this incredible technology lightweight and fun, because no one wants to think that it’s reading peoples faces, calculating credit scores, detecting lies and predicting insurance fraud as we go about our daily business. “When you say no to technology like that, you’re essentially saying no to having superpowers.” Fair enough you might say – but for people with autism, something as simple as reading a facial expression IS a superpower. It’s something their biology doesn’t give them for free. The standard technology for dealing with that today is cartoon faces on flash cards – but imagine being able to use the same technology used in lie detection to extract facial action code, and in real-time, teach kids how to better interact and read facial expressions. Using AI technology, these children are able to learn the theory of the mind, teaching them empathy and increasing their ability to learn why people do what they do. This form of augmented intelligence compensates for some of the things they’re not as good at and we often take for granted.
That same technology can also be used to identify objects in real time and communicate an audio description of any situation to someone who is visually impaired, providing them with an incredible amount of independence in being able to find things in the fridge, read streets signs, and find shops as they walk down the street, for example.
It’s also being used to help people who are mentally and physically impaired get better access to the government services and programs they need using digital humans created by Soul Machines, technology that is designed to read human emotion and respond accordingly. Instead of teaching people how to interact with computers – often a complicated and frustrating task for the disabled community, we’re teaching computers to interact with us in a more human way, creating a future where there is greater equality in access to information and digital services.
One example that I recommend you don’t try and Google (particularly at work) is Sexy Face – a game Dr Ming created for people to find everyone on Facebook they think is sexy, based on a system that has been designed to learn how people make judgements about faces. Allegedly for $5 they’ll find everyone that thinks you’re sexy too! This game was in fact a trojan built in partnership with Refugees United – designed to get people playing with the technology so it could learn to understand faces and get better over time, with the ultimate goal of being able to help find children in refugee camps. With over one million people in refugee camps around the world, you can imagine how difficult it is to find missing family members – particularly since the current method involves a book with one million photographs that are constantly changing! And yet with the technology created using Sexy Face, they were able to identify and find a child in any refugee camp around the world in just three minutes using an iPad. Talk about being able to change a life!
I don’t want to undervalue the use for AI in business – but we need to think about the impact of having (or not having) these technologies in society. If we don’t want visual recognition in the public, know that also means we won’t have it for autistic children, for people who are visually, mentally or physically impaired, or to help refugees find their families.
Or, potentially to help you be you for as long as possible. Navigating through space is one of the keys to preventing the onset of early dementia – so think about the impact that automated maps will have on your future and that of your children. Awesome technology, super convenient, couldn’t live without it – but fifty years from now there is a real risk we will see a significant increase in the onset of early dementia specifically due to the introduction of automated maps. And yet, AI is also being used to provide an early diagnosis and the augmented intelligence needed to minimise its impact. Dr Ming and her team have successfully been able to create a model of who you are – when you see a face you recognise, when you move through space, and your reactions to situations. As the disease progresses and you start to see a loss of function, think about what would happen in that first moment when you don’t recognise your own child – that terrible sense of loss of independence. The model can pick up the memory signature of that memory failure before you’re aware of it, and flash up your child’s name on Google glass prompting you to think “Of course that’s my son’s name!” and you don’t experience the memory failure in the first place. Most importantly it only turns on when you really need it – when it sees the memory failure coming – because good technology shouldn’t substitute what we do, it should make us better at doing it.
Need more examples? How about building a model on predictive coding in the retina to predict blood glucose levels hours into the future with a very high degree of accuracy, and sending real-time updates to the parents so they can monitor their child’s Type 1 diabetes from any where in the world, at any time. Or a model that predicts the onset of manic episodes in bipolar sufferers 3-4 weeks in advance, continuously looking for evidence of emotional change based on sensors attached and within their mobile phone. Or predicting life outcomes of young children and making recommendations on small tasks and activities that parents can do each day to change the course of their future and provide them a better quality life.
Keeping all these great examples in mind, there’s a few things we need to remember as we navigate our way through the fourth industrial revolution. First and foremost, AI is a tool that allows us to identify a real human problem, identify a real human solution, and make it available to everyone. When it comes to AI, we aren’t inventing things for the first time – and if we do, we aren’t creating solutions that are actual solutions. We can’t just throw technology at some data and expect a better outcome. We have to take the time to learn a subject – many of which have been studied for hundreds of years before us. It’s so easy to be misled if we don’t actually understand the problem domain and don’t actually understand the people involved. That’s the heart of AI – whether for good, or for business – we’re only ever solving human problems.
I wouldn’t be writing this if I didn’t truly believe AI can make the world a better place, but at the same time it doesn’t magically make it better. We have to think hard about the core problems we’re trying to solve – it’s not algorithms, it’s always people. When we keep that in mind, we’re not just making the world better for business, but also for our future and for that of our children.