We have work to do

While AI isn’t exactly new, its effects are truly starting to be felt in our everyday lives. And as with any new technology, it’s difficult for the public to trust it entirely, particularly when they don’t fully understand how it works.

The term artificial intelligence was coined by a Dartmouth College professor in 1956 and today refers to a range of techniques that enable computer systems to carry out tasks traditionally requiring human intelligence. With increasing complexity of technology, and many solutions built on techniques with arcane names like deep neural networks and reinforcement learning, it’s becoming harder and harder for the average person to understand how things work without specialised knowledge. We’re passing more analysis, decision making and problem solving responsibility over to computers and the vast difference between the hype surrounding AI and the reality of AI makes public perceptions on AI rather murky.

A recent survey conducted in the UK found that public understanding of AI tends to be quite general. Nearly 40% of respondents admitted their personal knowledge of AI was limited. Reactions to more specific characteristics of AI revealed that people tended to overestimate the level of sophistication and types of application AI systems are used for currently and will be in the near future.

This overestimation of AI is further evident when it comes to autonomous vehicles. Public perception around autonomous vehicles is dangerously positive. 70% of people believe you can buy fully autonomous vehicles today and 11% seem to accept the hype surrounding autonomous vehicles, saying they’d be willing to nap, read a newspaper, or watch a movie while using driver-assist features available today.

More worrying is people’s knowledge of how widespread AI currently is. 52% of respondents believed they had not yet been in contact with or used an AI application while 23% did not know whether they had or not. This leaves just 15% of the public who believe they have been in contact with or used an AI application.

Another interesting finding from the survey is the public’s perceptions on the use of personal data in AI systems. Many AI solutions, from online shopping recommendations to personal navigation systems, rely on personal data inputs to deliver smart, AI-derived outputs.  More than half of all respondents either thought AI would not use their personal data or did not know whether it would or not. Only 19% of survey respondents were okay with their personal data being used by AI solutions to perform tasks for them.

Since many current AI systems tend to be “hidden” within applications, product offerings or recommendations it’s tough for the public to learn more about how AI is really impacting their daily lives. The best way for the public to get used to and prepare for advances in new tech like AI is to see it in action and be educated about how it works and the benefits it would bring. As AI practitioners, it’s up to us to make sure solutions we build and that are used within the public domain, are coupled with material informing and educating people about how they function. Let’s avoid the hype and work to make our solutions easy to understand for everyone.