Autonomous cars in science fiction

See allHide authors and affiliations

Science Robotics  26 Feb 2020:
Vol. 5, Issue 39, eaax1737
DOI: 10.1126/scirobotics.aax1737


We view autonomous cars just like the ones in science fiction, and that could be a problem.

Radio-controlled automobiles were first demonstrated in 1925 (1), and only 5 years later, science fiction predicted full autonomy in Miles J. Breuer’s 1930 book Paradise and Iron. In Paradise and Iron, all of the cars, trucks, loading cranes, and boats on the mysterious island paradise of a wealthy inventor were self-driving. None of them housed a steering wheel; each could follow routes if given a map (reasonable given that GPS had not yet been invented), avoid obstacles, and call for help if stuck. They had all of the capabilities being proposed for modern-day autonomous vehicles except for a true natural language interface. These fictional autonomous vehicles were operating at level 5, the highest level of driving automation for on-road motor vehicles under the federally accepted SAE International standard J3016_201806 (2).

Of course, science fiction speculation about self-driving cars did not stop in 1930. Isaac Asimov predicted the public’s anxiety about the bounds of artificial intelligence in “Sally” (1950), a short story about an autonomous car that is more Stephen King than I, Robot. In Imperial Earth (1975), Arthur C. Clarke imagined a near future where it was illegal not to use self-driving when on the highway. In the 1980s, the popular TV series Knight Rider featured the most iconic autonomous car in fiction: KITT. KITT had all the impressive capabilities from Paradise and Iron combined with the natural language and self-awareness from “Sally,” resulting in a true personality. But KITT had to defer to its human driver, David Hasselhoff, unless Hasselhoff explicitly delegated control or was incapacitated. Most of the episodes played off the fact that KITT was often smarter than Hasselhoff’s character. In 1990, the movie Total Recall presented the opposite of KITT: Johnny Cab, the annoying autonomous taxi. Johnny Cab was the robot embodiment of the worst automated customer support center user experiences; it could only accept well-structured phrases in a specific sequence with no semantic understanding of what the passenger was saying. Fortunately, a human passenger, in this case, Arnold Schwarzenegger, could take direct control—if they were able to rip the Johnny Cab mannequin apart. Possibly the most imaginative vision of autonomous cars as a disruptive technology in science fiction to date has been from Vernor Vinge, the creator of the technological singularity concept. In his Hugo Award–winning novel Rainbows End (2006), individual car ownership is rare. Instead, a fleet of self-driving taxis, quickly and optimally, meets the transportation demands of the entire San Diego urban area.

After 2006, possibly due to the DARPA Grand Challenge, self-driving cars became less of a topic to be explored in science fiction and more of a world-building prop indicating that the book or movie was set in a believable future. The DARPA Grand Challenge, which ran from 2004 to 2005, and its follow-on, the DARPA Urban Challenge in 2007, captured the public’s attention with demonstrations of truly driverless cars. Although breakthroughs in self-driving cars were being seen in the late 1980s, the results were limited and were not being adopted and refined by industry or the military. The DARPA Grand Challenge changed the slow rate of evolution, and, with the intense press coverage, autonomous cars went from speculation to inevitable. Autonomous cars continue to appear in popular science fiction, such as Charlie Stross’ Halting State (2007) and Blade Runner 2049 (2017), but not as major plot points. There was no longer a need to speculate on how an autonomous car would work; technologists clearly knew the path ahead.

What science fiction as a whole missed, with the exception of Imperial Earth, were the legal and ethical implications of autonomous cars. The legal questions as to product liability and safely testing and evaluating cars on public streets are never addressed. Novels and stories introduce autonomous cars as being everywhere while avoiding how such technologies became ubiquitous. Science fiction has also ignored ethics. For example, KITT has features that, if incorporated into personal cars, could prevent driving deaths, but the inventor restricts the technology to a single individual performing vigilante crime fighting, implying that this is the greatest public good. This is rather like developing a cure for cancer but giving it to a single police officer rather than distributing it to the general public. In real life, there is something disturbing about how safety features in cars are expensive options that eventually work its way down to lower-cost models: Would it have been acceptable for only the children of rich families to get the polio vaccine for the first decade or so after its development? A recent article from the Brookings Institution analyzing state laws on self-driving cars noted that regulations for testing vary between states and that, despite two deaths in 2018, no state has modified their regulations in response to those deaths or has mechanisms for rapidly changing rules when data challenge underlying assumptions (3). It seems that self-driving vehicles are important enough for legislation but not important enough for legislation that might interfere with development (4).

Since the 1930s, science fiction has successfully predicted the capabilities now seen in autonomous cars; unfortunately, its silence on legal and ethical implications may be equally prescient. In Paradise and Iron, the value of self-driving vehicles was never questioned, possibly because the protagonists were too busy overthrowing the centralized machine intelligence that wanted to get off the island and take over the world. Eighty-nine years later, the value of autonomous cars for saving lives is unquestionable, but the conversation is still about what society should do to prevent a potential artificial intelligence uprising. More useful conversations might be about making sure everyone benefits from safety advances as soon as possible and how to find the right balance of regulations, capitalism, and research that will ensure safe experimentation while sustaining innovation.


View Abstract

Stay Connected to Science Robotics

Navigate This Article