There were doubts about driverless cars, robot-guided surgery, and even satellite navigation.
The view from the Royal Society? The AI industry needs to win over doubters. The practical suggestion was that AI will only be adopted when the benefits are clear to all. As it put it: “Society needs to give urgent consideration to the ways in which the benefits from machine learning can be shared across society.”
The survey has profound lessons for the financial industry. AI has incredible potential. It will improve the quality of financial advice. Lower the cost. Widen the audience. And be available round the clock, year-in-year out, in a way human advice simply can't match.
But gains will only be realised if users trust AI.
So how can we build trust? I think there are three ways.
1) Educate with actual use-cases.
AI is being used widely across the economy. It is widening access to service, and raising standards across the board. Let's shine a spotlight on the best examples.
Catch a flight and AI is making the grim process of check-in so much more bearable. An optical scanning engine reads your passport, opening the security gates. AI is used for predictive maintenance, so repairs can be done before aircraft break down. Security processes are streamlined with AI.
The entire airport process is improved with AI. Do passengers know this? They should.
Supermarkets are being turbo-charged with AI. The shelves are routinely stocked by AI engines. Morrisons supermarket uses deep-learning algorithms developed at the large hadron collider at CERN. The same algorithm used to discover the Higgs Boson particle helps Morrisons forecast future sales of carrots and courgettes to refine orders. More than three billion past transactions are crunched. Events such as weather predictions, and sporting contests are factored in by the AI engine. Then the orders automatically transmitted to suppliers. More than 200,000 items a month are purchased this way.
The old method, using humans with spreadsheets, was mind-numbingly dull, and inaccurate. How many shoppers realise that the full shelves come courtesy of AI? Again, not enough.
2) Talk about the benefits
Tech in the abstract can be baffling and boring. But explain the advantages and lightbulbs go on.
The health industry is supplying some terrific examples. An AI developed at Nottingham University can spot patients at risk from heart disease better than doctors. NHS England Professor Sir Bruce Keogh said, “We know from a number of studies that have been done that in certain circumstances AI is better than doctors at diagnosing certain conditions. It’s possible that certain types – and there are many different types of AI – will be able to read X-rays. I’ve been told by people who are developing this sort of stuff that’s within a four-year timeframe.”
The health secretary recently spoke of viewing GP interactions on an app “as simply as you can look at your entire order history on Amazon”.
AI helps consumers find the best deals. The energy industry is a nightmare for consumers. Comparison sites bring personalised advice to even the most technophobic consumer, saving them a fortune.
Explain the benefits, and the risks, and informed consumers will gravitate towards the solution that improves their lives.
3) Improve the Interface.
Smartphones didn't really catch on until the Apple iOS user interface, which used touchscreen icons. The old Windows Mobile phone menus were fiddly, and held back adoption.
Consumer-friendly UI's will bring AI to the masses, just as Dropbox brought cloud storage to the mass market by simplifying the interface. Consumers preferred Dropbox's simple design over the multitude of more complex and hard to fathom rivals.
Bank security is a field being re-invented by better interfaces. “Invisible” multi-factor authentication such as voice or typing speed are supplementing passwords, without consumers noticing. AI is behind both technologies. Data analytics reduced false positives by 60 per cent at Danska Bank. The results are enjoyed by users, without them being aware anything is happening.
The best example of all? Google's search bar. It hides a universe of AI technologies, yet it is so well designed the user is oblivious. That's why it's trusted with more than a trillion search requests a year.
The Royal Society is demanding the benefits of AI be spread as wide as possible. The AI industry can achieve this by meeting these three goals, thereby demystifying and unleashing the benefits of AI for all.