At certain points in prehistoric epochs, the evolution of life experienced an explosion of diversified organisms within a relatively short time. The so-called Cambrian explosion is the best-known of these diversification events. Scientists have debated the scope and possible causes for the explosion for decades. Fossil records show that between 505 and 542 million years ago, all of our existing biological taxa appeared, forever altering the global ecosystem.
AI and robotics systems may become pervasive in society in the same way. Hundreds of years in the future, historians could look back and pinpoint the “Cambrian explosion” of robotics: the moment when all types of artificially intelligent “beings” weaved their way into everyday life. Will this moment actually arrive? If it does, will it be a disruptive explosion or more like a simmering pot working up to a boil?
The Romanticism of a Moment
A large part of the Cambrian explosion’s appeal is the idea that there was some radical moment of upheaval that produced most of what we know as modern life. Within the space of about 35 million years, life evolved from microorganisms and soft-bodied creatures to animals with heads and tails, skeletons, shells and eventually eyes — a huge development that paved the way for modern biology. But many scientists argue that the existence of those soft-bodied Precambrian animals makes the Cambrian event less of an explosion and more of a significant lurch forward in evolution, probably caused by upticks in atmospheric oxygen or calcium, genetic bottlenecks or gene transfer.
The notions of a fully robotic future and a single explosive “spark” that sets it off are likewise romanticized. Sci-fi films and literature create worlds where humans and humanoid robots coexist in a master-slave or egalitarian relationship, but reality will likely be much more tempered than this. And truly disruptive technologies or moments in time make great news stories, but most revolutionary technologies require many moving parts to fully integrate into society. In the technological world, most innovations take decades or centuries to come to fruition. Truly disruptive innovations are extremely rare.
In a similar vein, many futurists like to predict a technological singularity — the explosive moment when artificial superintelligence will trigger runaway innovation that will change (or possibly exterminate) humanity forever. Technologists like Ray Kurzweil have gone so far as to predict a date for this event (2045, in his case), but many researchers criticize the movement for its almost religious fervor in support of a highly speculative event. It’s fun (and often terrifying) to think of these long-simmering developments building to an explosive moment, but it’s also unrealistic.
Realistic Projections and Sensible Timeframes
If artificially intelligent robots don’t suddenly integrate into society in an explosive moment, how will the robotics revolution come to pass? The answer depends on dozens of factors, both technological and societal.
A Future That Works: Automation, Employment, and Productivity — a comprehensive McKinsey Global Institute report published last year — provides some insight into the question. The report lists five factors that affect the pace of widespread automation: technical feasibility, development and deployment costs, market dynamics, economic benefits and social acceptance. The authors then use these factors to model four stages of adoption: technical potential, solution development, economic feasibility and finally, adoption.
The report goes on to examine the current state of automation technology, assuming that an artificially intelligent robot could perform work at the same level as a human agent when it achieves roughly human capabilities (see report, page 72). It found that while current automation systems are closing in on human performance in the areas of coordination between agents, social sensing and dexterity, they have a long way to go as far as creativity, emotional reasoning and natural language processing. Examining even a single factor like technical feasibility shows that fully adopted automation is a slowly evolving process.
Even after widespread automation is found to be technically, economically and socially feasible, it may still have decades to go before it becomes ubiquitous. Consider the history of flight, which began with isolated experiments and devices over several centuries. Balloons became possible in the 18th century, leading to the development of airships. After centuries of “heavier-than-air” experiments, the Wright brothers successfully flew in 1903. The first commercial flight was established in Florida in 1914, and widespread commercial aviation started around 1926. Commercial flight saw a boost following World War II, and it took until the 1970s to see widebody jets establish the current standard of air travel.
In step with the development of foundational technologies like air travel, the McKinsey report found that adoption rates range from eight to 28 years for most technologies. It also found that automation — which promises to transform the world of work and labor — may need more time, well beyond its technical development, to fully integrate into multiple industries’ processes, procedures and regulations. For many technologies, decades could pass between the point of first adoption to full adoption.
Barriers to AI Acceptance
Even if automation and ubiquitous robotics is at some point fully ready for adoption, it will likely have social barriers to breach as well. While societies like Japan openly embrace many robots, most of the world treats them more skeptically. But here it’s useful to distinguish between robot types — like biological life itself, robots are a diverse group with varied forms, tasks and sizes, among other attributes. An American or European may be completely accepting of an artificially intelligent manufacturing process but totally averse to social or caregiver robots.
Like the environmental factors governing life in Cambrian seas, these cultural attitudes will likely play a part in robot diversity in the future. A completely humanoid robot that frightens 95 percent of the populace will never be adopted on a wide scale, despite its utility and technical feasibility.
But it also makes sense to ask: is the robotic Cambrian explosion already taking place? Consider the popularity of Roomba vacuums, intelligent personal assistants, self-driving vehicles and more. Is the world witnessing a diversified robot evolution?
According to Gartner’s 2017 Hype Cycle for Emerging Technologies, smart robots are nearing the “peak of inflated expectations,” after which point they will (theoretically) tumble into the “trough of disillusionment” before leveling off to a productive plateau of general acceptance and adoption. Perhaps the fascination with a fully diversified robotic future is part of the inflated expectations and the automated future will remain similar to the present. Or perhaps the coming decades will see the rise of robot servants, and maybe even robot uprisings — a far-fetched but fully possible scenario that’s made for some great science fiction.