Future of autonomous vehicles depends on driver attitudes
Getting the public behind the idea of an autonomous vehicle means peeling back the black box nature of AI and proving the safety of self-driving technology.
Though often pegged as the future of transportation, the demand for a fully autonomous vehicle remains tempered by user uncertainty.
Consumers are hesitant about handing over the steering wheel to AI. Semi-autonomous cars have shown tremendous ability to keep users safe in challenging road conditions but have not won over most of the public. Despite the great potential of autonomous cars, there remains a broad feeling of distrust in the average consumer.
Autonomous vehicles and the public
Myplanet, a Toronto-based software company that builds user interface technology recently conducted a survey of 3,000 consumers. It found that only 15% of respondents were comfortable with autonomous driving. Myplanet's research reinforced an undercurrent of distrust between the public and the automated vehicle.
"Autonomous driving is ranked amongst the most undesirable across all types of artificial intelligence and machine learning technologies," Jason Cottrell, CEO of Myplanet said. "However, automotive task-oriented technology or assistive AI ranked far more favorably."
The lack of trust the public has is linked to the high expectations that the average driver has when it comes to vehicle safety. There have been a few public instances of catastrophic autonomous driving failures that have soured the public.
Principle analyst at Forrester Michele Pelino sees the public perception of failure as a major catalyst for consumer anxiety around autonomous vehicles.
Humans recognize the small chance of driver mistakes and live with them, but driverless vehicles ask the user to relinquish the control over the steering wheel to the AI. The autonomous car cannot afford to make a mistake because the risks are too high.
To build that trust, the automotive industry and builders of semi-autonomous cars need to be near-perfect, trustworthy and transparent.
The path to adoption and acceptance
Organizations seeking to further the autonomous vehicle need to focus on proving their technology through micro-automations. Automated vehicles are unproven on a grand scale -- where a digital agent is operating the entire vehicle -- but automated processes in cars are becoming more common and trusted in the general public. Through the proliferation of brake-assist technology and automatic parking, the idea of giving away control to an AI becomes more palatable.
In Myplanet's two-part survey of people across the United States, 56% of respondents indicated that they were comfortable with automated technology such as lane departure assistance and 47% with automatic park assist. These small iterations of autonomous technology are where trust is earned.
Task-oriented driver aids are the path to proving the utility of autonomous vehicles, Cottrell said.
By augmenting drivers, rather than replacing them, the automotive industry can take a large step forward in earning the trust of the wider market. Automated parking is a key opportunity for this growth of trust; people can see autonomous technology improving safety and reducing the guesswork of trivial tasks such as parallel parking.
"Similarly, 'autopilot' respondents, while still wary, are far more comfortable when offered some of the functionalities of autonomous vehicles when [they are] presented as assisting humans and task control," Cottrell said.
The framing of technology is important when it comes to automated driving because of the risks involved with AI's decision-making process. When humans are driving, they can be flexible and adaptive. Most importantly, they understand the decisions they make. A driverless car can make an incredible amount of observations and decisions in a tiny amount of time but all without the passenger's knowledge.
Michele PelinoPrinciple analyst at Forrester
"You must frame [autonomous driving] in a way that consumers understand what has been thought through, what has contributed to the decisions that are made and how quickly those decisions are made," Pelino said.
The black box nature of AI leaves the general public uncomfortable and confused. To reach a wider audience, the driving system needs to be relatively explainable and its decisions accounted for through white box testing.
AI and perception
There's still a general discomfort around AI and automation across enterprises, and the problems that prevent the wider public from accepting semi-autonomous vehicles are the same that other forms of AI face. The push towards advanced artificial intelligence has been matched by people unwilling to buy-in as the result of safety concerns, fears over an autonomous workforce or limited ROI.
Myplanet survey respondents said they were more comfortable with more task-oriented AI and automation including technology such as hotel check-in kiosks, lane departure assistance and home automation.
By revealing the ways in which automated vehicles operate and emphasizing the augmentation capabilities of AI, organizations and companies can aim for a larger market and build trust over time.
"They are putting those things in [current car technology] not specifically saying 'autonomous driving,' but as value-added service capabilities and features that the car has," Pelino said.
For now, the public remains resistant to a fully autonomous vehicle, but with more augmentative technologies being deployed in the auto industry, the driverless vehicle will eventually win over the wider public.