Rodney Brooks knows the difference between true technological advances and unfounded hype.
One of the world’s most accomplished experts in robotics and artificial intelligence, Brooks is the co-founder of IRobot, maker of the Roomba robotic vacuum cleaner. the co-founder and chief technology officer of RobustAI, which makes robots for factories and warehouses; He is the former director of the Computer and Artificial Intelligence Laboratories at the Massachusetts Institute of Technology.
So when, in 2018, Australian-born Brooks encountered a wave of unwarranted optimism about self-driving cars — “People were saying outrageous things, like, Oh, my teenage son will never have to learn to drive” — he took it as a personal challenge. In response, he’s compiled a list of predictions about self-driving vehicles, artificial intelligence, robotics, and space travel, and promises to review them every year until January 1, 2050, when he’ll be 95 if he’s still alive.
I don’t think we’re limited in our ability to build humanoid robots, after all. But whether we have any idea how to do it now or if all the methods we think will work are remotely correct is entirely up for grabs.
Robotics and artificial intelligence expert Rodney Brooks
His goal was to “inject some reality into what I saw as an irrational exuberance.”
Each prediction carries a time frame – maybe something happened on a certain date, not before a certain date, or “not in my life”.
Get the latest from Michael Hiltzik
Commentary on Economics and More from a Pulitzer Prize Winner.
You may occasionally receive promotional content from the Los Angeles Times.
Brooks posted his fifth annual scorecard on New Year’s Day. The majority of his predictions were spot on, though this time he admitted he thought he, too, had allowed the hype to make him overly optimistic about some developments.
“My current belief,” he wrote this year, “is that things will go, on the whole, more slowly than I thought five years ago.”
As a veteran technologist, Brooks has insights into what makes ordinary people, or even experts, overly optimistic about new technologies.
People have been “trained by Moore’s Law,” Brooks told me, to expect that technologies will continue to improve at ever faster rates.
His reference is to an observation made in 1965 by semiconductor engineer Gordon Moore that the number of transistors that could be fitted on a microchip doubled approximately every two years. Moore’s observation became a proxy for the idea that computing power will improve exponentially over time.
This tempts people, even experts, to underestimate the difficulty of reaching a chosen goal, whether it be self-driving cars, self-aware robots, or living on Mars.
He told me, “They don’t understand how hard it is to get there, so they assume it’s just going to keep getting better.”
One such example is self-driving cars, a technology with limitations that ordinary people rarely recognize.
Brooks wrote about his experience with Cruise, a service that uses self-driving taxis (with no one ever in the front seat) in parts of San Francisco, Phoenix, and Austin, Texas.
In San Francisco, Cruise only operates between 10pm and 5:30am—that is, when traffic is lighter—and only in limited parts of the city and in good weather.
On his three cruises, Brooks found that vehicles avoided left turns, preferring to make three right turns around a block instead, driving very slowly and once trying to carry him in front of a construction site that would have exposed him to oncoming traffic.
“The result is that it was two times slower than any human-operated transportation service,” Brooks wrote. “It may work in specific geographic areas, but it won’t compete with human-run systems for a long time.” He also said that it is “decades away from profitability”. In his annual scorecard this year, he predicted that “there will be human drivers on our roads for decades to come.”
The annual scorecard is one of several outlets Brooks relies on to mitigate the “irrational exuberance” around technology in general and artificial intelligence in particular. He has been a frequent contributor to IEEE Spectrum, the home member of the leading professional society for electronics engineers.
In an article titled “An Inconvenient Truth About Artificial Intelligence” in September 2021, for example, he noted that each wave of new developments in AI has been accompanied by “breathless predictions about the end of human domination of intelligence” amid a “tsunami of promise, hype profitable applications.
In fact, Brooks writes, nearly every successful deployment of AI in the real world has either had a human “somewhere in the loop” or a very low cost of failure. He wrote that the Roomba operates autonomously, but that its more serious failure would involve “losing a piece of land and failing to catch a dust ball”.
When IRobots were deployed to Afghanistan and Iraq to disable improvised explosive devices, “a failure there could kill someone, so there was always a human in the loop giving supervisory orders.”
Robots are common today in industry and even around the home, but their capabilities are very limited. Robotic hands with human-like dexterity haven’t advanced much in 40 years, Brooks says. This also applies to independent movement around any home with clutter, furniture and moving objects. “What is easy for humans is still very, very difficult for robots,” he writes.
For ChatGPT, the creator of AI prose that has garnered a lot of attention from high-tech enthusiasts, along with warnings that it could usher in a new era of machine-driven plagiarism and academic forgery, Brooks argues for caution.
“People are making the same mistake they’ve been making over and over,” he wrote on his scorecard, completely mistaking some new AI demo as a sign that everything in the world has changed. did not happen “.
He writes that ChatGPT repeats patterns in a human prompt, rather than showing any new level of intelligence.
This is not to say that Brooks doubts the eventual creation of “truly artificial intelligence, with cognition and consciousness distinctly similar to our own,” he wrote in 2008.
He predicts that “the robots that will roam our homes and workplaces…will emerge gradually and symmetrically with our society” even as “a wide range of advanced sensory devices and prosthetics” emerge to enhance and strengthen our bodies: “As our devices become more like us, we will become more like them.” And I am optimistic. I think we’ll all get along.”
This brings us back to Brooks’ scorecard for 2023. This year, 14 of his original predictions were deemed accurate, whether because they occurred in the time frame he predicted, or because they failed to happen before his deadline.
Among them are driverless package delivery services in a major US city, which he predicted won’t happen before 2023; It hasn’t happened yet. In terms of space travel and space tourism, expect a suborbital launch for humans by a private company to happen by 2018; Virgin Atlantic beat the deadline with such a flight on December 13, 2018.
He predicted that spaceflight with a handful of paying customers wouldn’t happen before 2020; regular flights no more than once a week no earlier than 2022 (possibly by 2026); and fly two paying customers around the moon no later than 2020.
All those deadlines have passed, which makes predictions accurate. Only three flights took place with paying customers in 2022, which indicates that there is “a long way to go to get to the sub-weekly flights,” notes Brooks.
Brooks constantly questions the predictions of the most-cited tech entrepreneur, Elon Musk, who Brooks notes “has a pattern of overly optimistic time-frame projections”.
Brooks notes that lunar orbit for customers pushing in Musk’s SpaceX Falcon Heavy capsule doesn’t seem possible before 2024. Landing the payload on Mars for later use by humans, which Musk predicted would happen by 2022, doesn’t seem to happen before 2026, and even This date is “overly optimistic”.
Musk didn’t deliver on his 2019 promise that Tesla would put a million automated taxis on the road by 2020 — that is, a fleet of self-driving cars called through a Tesla Uber-like app. “I think the actual number is still firmly zero,” Brooks wrote.
As for Musk’s dream of regular service between two cities on his Hyperloop underground transit system, Brooks puts that in the “not in my life” hole.
Many of Brooks’ predictions remain open, including some relating to the electric vehicle market. In his original forecast, he predicted that electric vehicles would not reach 30% of US auto sales before 2027 or 100% before 2038.
Growth in electric vehicle sales becomes turbocharged in 2022 – increasing 68% in the third quarter over the same quarter a year earlier. If this growth rate continues, electric vehicles will account for 28% of new car sales in 2025.
This assumes that the driving forces for EV adoption continue. Head wind, however, should not be underestimated. Electric vehicle sales may have spiked due to the massive hike in gasoline prices in 2021 and last year, but that inflationary trend has now disappeared. Battery plants may take longer to come online than expected, which could lead to shortages of these critical components and drive up electric vehicle prices.
“There is clearly something going on,” Brooks wrote, though “the jury is still out” on whether the US will see 30% market share for electric vehicles by 2027.
Brooks does not wish to stifle human aspirations to build robots, artificial intelligence systems, or space exploration.
He told me “I’m a technician”. “I build robots — that’s what I’ve done with my life — and I’ve been a space fan forever. But I don’t think it helps people to be so overly optimistic off the charts” that they ignore difficult problems that stand in the way of progress.
“I don’t think we’re limited in our ability to build humanoid robots, eventually,” he says. “But whether we have any idea how to do it at the moment or whether all of the methods we think will work are just right is entirely up to you.”
The dream is compared to the dream of medieval alchemists searching for how to turn lead into gold. “You can do that now with a particle accelerator to change atomic structures, but at the time they didn’t even know atomic structure existed. We might as well be at the level of human intelligence, but we have no idea how it works at all.”