Robo-advisors that are indistinguishable from their human counterparts are no more than three years away, estimates one fintech expert.

By 2020 robo-advisors should be able to pass the Turning Test, David Horton, global head of digital innovation at Synechron told Markets Media.

The artificial intelligence test, which British mathematician and computer scientist Alan Turning devised in 1950, defines artificial intelligence as successful if a person cannot distinguish between the answers given by someone else and an AI engine.

“All you have to do is have the robo-advisor go through a machine-learning algorithm to understand the regulatory conditions and the boundaries of risk that it likely will find and the robo-advisors will be able to execute against those very effectively,” said Horton. “There are always examples where robo-advisory solutions are out-performing their human counterparts.”

However, early AI adoption has been more in the customer service space than the investment space he added.

Horton does not find this that surprising since the consumer market has been the first industry vertical to set the bar for user expectations with products for the Internet of Things like Apple’s Siri, Amazon.com’s Echo, and Google’s Google Home.

Of the more than 200 senior-level, global financial services business and IT decision-makers across the U.S. U.K. and Europe surveyed by industry-research firm Tabb Group on behalf of Synechron, approximately one in ten cited that AI was a top priority for them in 2017.

When grading the sector’s adoption of AI, Horton gives the leading firms a ‘B’ and the majority of businesses a ‘C’ or a ‘D.’

The grade disparity is due to competition for the limited amount of experts in AI development and training, which is similar to the industry’s experience with other nascent technologies like blockchain, he explained. “It takes a little bit of time for the industry to show it has significant investments to make before that skill set ramps up and you have much more talent available.”

As more and more enterprises implement AI-based platforms, their adoption will drive down the price of the technology further.

“Some of the AI solutions like IBM Watson is considerably more affordable today that it was many years ago,” said Horton. “That’s not to say that Watson is cheap, but the factor that it would have cost 10 years ago would have been a lot more expensive.”

Since AI technology matures incrementally on a daily basis and it still is in the early part of its evolutionary cycle, it is natural that everyone will graduate and become a little better at implementing AI technology in the next couple of year, he added. “I would say by 2020 we are likely to see a lot of impressive artificial intelligence available through product and services offered by companies.”