To Automate Personalisation We Must Automate Empathy
People pay for personal. From the shop-floor to the driving seat, we crave authenticity and uniqueness: craft beer, upcycled furniture, custom paint-jobs. In the digital world there is potential for unlimited customisation, with only marginal cost for mass-scalability. But designing for one can be a trade-off against the other. Now even the weakest AI can make digital systems noticeably smarter, paving the way for personalisation at scale and a new level of enchanting experiences. But to tune themselves ever more to our needs, the machines need to know more about us. With the emergence of new sensors and algorithms, we can now measure user emotions for even richer personalisation. Better human experience requires deeper human understanding.
The value we place on personalisation is hard to fathom. Back in 2009, when Netflix gave a $1M cash prize to a team that improved its movie ratings system by just 10%, they sent out a clear message: personalisation means profit. Similarly, Amazon understood early on that it would need to embark on an unprecedented data-mining expedition just to begin to make useful product recommendations. And let’s face it, the recommendations of Netflix, Amazon and other leaders are still frequently disappointing. Perhaps for some, big data is never big enough.
Today we are seeing the development of personalised transport, robots, voice-controlled assistants, smart homes, medicine... All our objects are being connecting to the internet and to each other, and in the process being enhanced by a nascent ecosystem of AI-fuelled data processing. Sensors for automatically measuring us humans as we interact with all these devices and services have been dropping in price rapidly, while simultaneously rising in variety, quality and number. For instance, there are vehicles in the current production cycle that are being fitted with hundreds of sensors for understanding the current state of both the machine and its occupants on a microsecond basis. These connected vehicles could stream hundreds of gigabytes of information per hour into the network, where it can be analysed by smart software to make decisions on our behalf, hopefully saving lives and making our journeys more efficient and enjoyable.
Central to any successful product, service or experience is the belief that we should design everything with the user in mind. Whether you assign it a trendy name like ‘human-centred design’ or an old one like ‘the customer is always right’, it’s not a new idea. We’ve had a name for this for over a century: empathy. You have to understand people, and how they feel, if you want to make things that delight them.
We have become used to the idea of digital systems knowing a lot about us. We provide demographic information when we create a personal profile for a new service, and we allow activity tracking to monitor what we’re up to through tools such as cookies. The next evolutionary stage for mass-personalisation is to go beyond knowing who the user is, or tracking what they do, and finally understanding how they feel. Welcome to the age of empathic technology.
Companies like ours are working to improve digital systems so they understand what the user is feeling and respond appropriately. Largely through the use of sensors that measure physiological changes like biometrics, facial expressions and voice tone, empathic tech aims to know how you are not just what you are. Then, to extend that human understanding up to large scales, we need smart automation, running algorithms that are trained to understand and respond to the user’s emotions. In other words, empathic AI.
If this sounds far-fetched, creepy or banal, it’s probably because the technology is still at an early stage. It’s noticeably new. Nobody yet knows where it will go, what the ‘killer apps’ will be or who will abuse it. Soon, I suspect, it will work without us even noticing, because innovations tend to become familiar and disappear into the background of a technology experience.
‘Any sufficiently advanced technology is indistinguishable from magic’.
– Arthur C. Clarke
‘As soon as it works, no one calls it AI any more’.
– John McCarthy
Of course, invisibility is no excuse for ignorance. Our trust in data-hungry organisations has been severely dented by one hacking or surveillance scandal after the next. Not everyone will stick to Rule No. 1: don’t be a dick. And regulatory measures like GDPR are only partial, they tend to show up to the party late, and can be misused as excuses for offloading responsibility to the user by ‘informing’ them.
Anyone working with empathic tech needs to think hard about how to protect and respect their users from the outset. Against the potential value of delightful hyper-personalisation we must consider the possible social costs we could be risking for mistakes, bad decisions or outright exploitation. All service-providers, from charities to corporations to governments, have a duty to engender new levels of trust in future generations. And, at least in the mind of each customer, user or citizen, that gift of trust is given not just to the organisation providing the service but also in an increasingly complex interaction directly with our ever-smarter machines. So we must be able to trust the machine not to be a dick, too.
At Sensum, we look at smart systems and ask how they can be made to think and act in more human ways, by bringing user emotions into the mix. How ‘smart’ a digital system can be is partly limited by its capability to sense, understand and respond to the user’s mood, situation or environment. For a familiar example, take music. A service like Spotify places enormous value on recommending what you might want to hear next (and I think they do an exceptionally good job of it). Imagine how that recommendation could then be improved if the service knew how you felt while listening to each tune. Taking that a step further, it could learn to predict your particular desire to either amplify or attenuate your mood. When you’ve had a stressful day, maybe you shake it off with heavy metal, or wind down with deep chill. Or both, depending on circumstances.
Everyone is different in subtle little ways. Teams like ours are dedicated to training digital systems not only to understand how each of us feels from one moment to the next, but also to have the artificial emotional intelligence to respond appropriately. This will improve with more data, research, ingenuity and conscientiousness, leading to a new generation of empathic digital experiences tailored to the individual. Personalisation needs empathy, and now empathy is being automated.
Main image credit: Josh Sorenson (via Pexels)