AI gets under the skin at CES by Arthur Goldstuck

“AI is no longer a ‘thing’ – it’s everything,” were the breathless words from the Consumer Technology Association, organisers of the annual Consumer Electronics Show in Las Vegas, when it introduced an artificial intelligence conference track at the event last week.

As much as it sounds like hype, one could take the claim a step further, and say “AI is IN everything”. In the same way that last year’s CES was brimming with voice assisted technologies, one could hardly move at this year’s edition without stumbling across a new TV set or gadget that claimed to have AI built in. One of the drivers of the trend is the arrival, en masse, of the Internet of Things, which refers to the ability of connected devices to be networked via the Internet and interact or exchange information. For vast numbers of such devices to operate effectively, AI has become an essential underlying technology.

The CTA summed it up: “From workplace automation, biometrics and machine learning to security, accessibility and enhanced robotics, AI is, today, integrated into virtually every piece of IoT technology; a trend that will grow exponentially as we approach a 5G world.”

Last year, 5G, the next generation of mobile connectivity, was also hyped to the hilt, whereas this year it, too, retreated beneath the skin of gadgetry.

The biggest challenge for AI is one that is as invisible as the technology itself: the ethics of allowing software to make decisions that affect humans. The many debacles facing social networks like Facebook and Twitter are a result of allowing AI, in the form of algorithms, to access sensitive personal information and use it to target advertising and political messaging.

At CES, AI was demonstrated in action with facial, gesture and voice recognition. It allows robots, for example, to become more responsive to their owners. One of the more ludicrous examples was a family companion robot called Lovot, which has little purpose other than using its imploring cartoon eyes to get humans to pick it up and cuddle it. Yet, this toy has more than 50 sensors and cameras built in to interact with its environment. Other toy robots are programmed to respond only to family members and to raise the alarm if they sense strangers, using AI.

In other words, the most expensive electronic toys of today are really test beds for the technology of tomorrow. But there is little evidence that AI is being developed with a focus on winning the trust of humans. Facial recognition, for example, has been shown to be weak at identifying non-Caucasians, a result of being trained with the faces of white people. The CTA addressed the issue in a conference session entitled, “Solving bias in artificial intelligence”, but the message was hard to find on the show floor.

IBM CEO Ginni Rometty, in the CES opening keynote address, said that AI would prove that data is the “world’s greatest natural resource,” enabling revolutions from smart cities to health care, transportation to robotics. The subtext, however was that humans of the early 21^st century are not proving themselves especially responsible in looking after the world’s natural resources.