Resolving the Google Steering Wheel Dilemma


30 October 2016

Smart machines will be a top five investment priority for more than 30 per cent of CIOs by 2020, according to IT research and advisory company Gartner.

However, with smart machines moving towards fully autonomous operation for the first time, balancing the need to exercise control versus the drive to realise benefits is crucial, said Gartner.

Speaking at the Gartner Symposium/ITxpo in the Gold Coast, Brian Prentice, research vice president at Gartner, said Google’s self-driving car project is a perfect example of why pursuing full autonomy may be neither possible nor desirable in smart machines.

Human beings are still required as the final point of redundancy in an autonomous vehicle, so a fully autonomous car requires a steering wheel should a driver be required to take control, he said.

“But putting a steering wheel in an autonomous car means a fully licensed, sober driver must always be in the car and prepared to take control if necessary." He added, "Not only does this destroy many of the stated benefits of autonomous vehicles, but it changes the role of the driver from actively controlling the car to passively monitoring it for potential failure."

Gartner believes the “Google Steering Wheel Dilemma" is representative of a challenge all smart machine initiatives must face.

Smart machines respond to their environment, Prentice said. “But what is the environment that the smart machine is responding to? Environments that are largely uncontrollable are not amenable to smart machine projects because it is difficult, if not impossible, to model accurately.”

The trick then, he said, “is to figure out what is actually controllable and limit smart machines to that which can be accurately modelled and managed”.

According to Gartner, CIOs seeking to maximise the benefits of smart machine solutions must:

· Plan to deliver smart machine-enabled services that assist and are overseen by humans to achieve maximum benefit in the next three to five years, rather than those that are fully autonomous.

· Identify and analyse the constraints within the environment that the eventual solution will face at the beginning of any project aiming to make use of smart machine technologies.

· Design any smart machine solution outward from constraints identified in the key areas of user experience, information asymmetry and the business model to hit the sweet spot for smart machine-enabled solutions, and maximise the benefit the technology will provide.

Prentice also stressed that major unresolved problems in machine learning solutions, such as how to ensure learning data is fully representative and how to avoid "reward hacking" need to be addressed before any autonomous machine that continues to learn from its environment can be deployed as a mass-market solution to a real-world problem.