Emotionally aware AI is imminent, says futurist Shivvy Jervis

An AI avatar voiced by Cate Blanchett that can read emotions is being rolled out right now by a major bank

Pretty soon we will have to get used to having our faces regularly scrutinised in minute detail. Like a suspicious lover, alogorithms will be watching for the telltale, barely susceptible micromovements that betray our emotions. And it will all be in the cause of providing us with a better digital experience.

User experience is now the differentiator that trumps all others, including price, futurist Shivvy Jervis told the audience at the Computing IT Leaders Summit on Tuesday. And increasingly, UX depends on understanding us intimately as individuals.

An adviser on new technology to governments and enterprises, Jervis, who was once head of digital media at Telefonica, said the title of 'futurist' was not one she'd chosen, but she'd since gladly adopted it in part because of the lack of female representation in the field. Currently she's involved in a project for the Mayor of London, examining what the city will look like in 25 years. "I'm not interested in the short-term fads, I'm interested in the technologies that I think will go the distance," she said.

One of these is conversational commerce, where every barrier to interaction between humans and machines is being smoothed away, leaving an interface that's frictionless, always available, takes the fewest possible steps to arrive at the end goal and can be rolled out at a low cost. Increasingly, the technology is being developed with human need as the starting point, rather than requiring people to adapt to it, as is so often the case.

Emotionally aware AI

The public faces of conversational commerce are hyper-realistic avatars, their AI-derived looks selected for broad appeal. These avatars will soon be the first point of contact that we have with many organisations. Like today's chatbots they will answer our most frequently asked queries, passing us to a human assistant if necessary. Unlike simple chatbots though, they will adapt their responses in accordance with our emotions, as revealed by our smartphone cameras or through voice technology. This will make for a far more naturalistic conversation between human and machine. Also unlike most chatbots, they will both feed and plug into a knowledge base derived from machine learning so that before long they will be able to answer almost every question in a contextually appropriate manner.

Such systems are already being developed by the firm Soul Machines and others. Voiced by Cate Blanchett, one such avatar is due to go live within weeks at a major US bank while two governments are rolling out their own versions.

"It is emotionally aware AI," Jervis said. "It still looks a little stilted but it's nearly there."

Emotional scanning is finding its way into cars too, where dashboard-mounted cameras by companies like Veoneer can monitor drivers' faces to assess 'cognitive load', watching out for the momentary losses of concentration that can lead to accidents, and alerting the driver to take care.

It all comes back to data

Such helper systems are evolving at the intersection of a number of trends. In addition to the focus on UX there is personalisation, with every individual treated as a unique being on a unique journey (personalised medicine being a well-known application). This in turn is made possible by predictive analytics which is enabled by data scientists being able to train models with large datasets.

In the end it all comes back to data. However, most organisations still haven't got around to putting even rudimentary structures in place to extract value from their data. The analytics divide is as wide as it ever was and it will grow wider as data-driven intelligent systems go mainstream. "We need to bring the tools that were previously only in the toolbox of data scientists into our organisations, using data to establish patterns that you and I cannot hope to see," Jervis said.

Just as the boundaries between human and machine are becoming increasingly blurred, so the real and virtual worlds are intermingling in the form of mixed reality. Largely the preserve of gaming (think Pokemon Go and derivatives), retailers such as IKEA are already offering apps that allow you to see how an item of furniture might look in your home, and such technologies will become increasingly common in the workplace too, Jervis said.

Make way for the human-bot interaction counsellor

It's hardly news that machines are already "taking the robot out of the human", automating routine work and leaving more time for creative, intuitive and interpersonal tasks, but what's less well understood is the role of so called 'IQ bots' that can intelligently surface relevant information from screeds of unstructured data in order to help people make nuanced and informed decisions. These intelligent agents will become virtual members of staff, said Jervis, although most of us shouldn't fear for our jobs. It will be a long time before machines can negotiate a contract, deliver a motivational talk or offer a listening ear.

"Humans will increasingly be interacting with bots, but they'll start to take the tedium out of our everyday jobs, augmenting our work rather than displacing it."

As new virtual helpers come into the workforce, novel roles will be required to management them. Jervis laid out a few speculative job titles of the near very future. These include a head of immersive workplace, a chief trust officer to establish the integrity of a company's information and (a little way off) a human-bot interaction counsellor.

"I'm not sure when we'll see that one, but I do believe we'll need someone like that in the near future," she said.