A couple of weeks ago I attended MIT’s Second Machine Age Conference, an event inspired by the best-selling book of the same title published earlier this year by MIT’s Erik Brynjolfsson and Andy McAfee. The conference presented some of the leading-edge research that’s ushering the emerging second machine age, and explored its impact on the economy and society. It was quite an interesting event. Let me discuss a few of the presentations as well as my overall impressions.
In his opening keynote, Brynjolfsson explained what the second machine age is all about. “Like steam power and electricity before it, the explosion of digitally enabled technologies is radically transforming the landscape of human endeavor. Astonishing progress in robotics, automation, and access to information presents major challenges for institutions from small businesses and communities to large corporations and governments, but it also creates opportunities to rethink how we live and work in profoundly positive ways.”
The machines of the industrial economy, - the first age, - made up for our physical limitations, - steam engines enhanced our physical power, railroads and cars helped us go faster, and airplanes gave us the ability to fly. For the most part, they complemented, rather than replaced humans. The second age machines are now enhancing our cognitive powers, giving us the ability to process vast amounts of information and make ever more complex decisions. They’re being increasingly applied to activities requiring intelligence and cognitive capabilities that not long ago were viewed as the exclusive domain of humans. Will these second age machines complement or replace humans?
Brynjolfsson said that he’s amazed at the advances in machine intelligence in the last decade. They’re able to sense and interact with the physical world, like the Google self-driving cars. Their vision and fine motor control has significantly advanced, like Baxter, the interactive production robot from Rethink Robotics. We’re now able to communicate in natural language with smartphone apps and customer service applications. And question-answering systems like IBM’s Watson are enabling us to solve increasingly complex problems.
“We’re in the midst of the greatest one-time event in history!,” he added. But, what does this mean for the economy? On the one hand, we have a bounty of highly sophisticated and inexpensive technologies, as exemplified by a Radio Shack ad from the 1980s, where the functions of just about every single one of the devices then on sale are now available in our smartphones, - computer, phone, messages answering, music, video and audio recorders, and so on.
GDP and profits have significantly increased over the past few decades. Jobs, however, are down. This has led to what Brynjolfsson and McAffee called The Great Decoupling in a December, 2012 NY Times OpEd.
In the second half of the 20th century, productivity and GDP growth led to increased job creation, higher incomes and a rising overall standard of living. But, this link now seems to be broken. While productivity and GDP have continued to grow, employment and income have been declining for the past decade. Millions have been left behind, because many organizations and individuals lack the skills to keep up with the rapidly advancing technologies. “Digital progress makes the economic pie bigger,” noted Brynjolfsson. “But there is no economic law that everyone, or even most people, will benefit.”
He finished his keynote by summarizing what he called The New Grand Challenge: Digital technologies will continue to accelerate; Our skills, organizations and institutions are lagging; Business as usual won’t solve this problem. “We need to reinvent our economy and society to keep up with accelerating technologies.” To help do so, Brynjolfsson and McAfee have co-founded and co-direct the new MIT Initiative on the Digital Economy.
Subsequent presentations gave us a good overview of some of these accelerating technologies. Daniela Rus, MIT professor and Director of the Computer Science and AI Lab (CSAIL), talked about One Robot for Every Task. “The digitization of practically everything coupled with the mobile Internet, automation, and advanced robotics promises a future with democratized use of machines and wide-spread customization. Increasingly, cognitive machines and technologies for rapidly designing and fabricating new, more capable robots have the potential for broad-based impact, from manufacturing and construction to health-care, smart cities, transportation and other facets of everyday life.”
In a recent interview, Rus said that she expects robots to be as commonplace as smartphones. This will include simple, inexpensive self-assembling robots, some of which can be printed in less than an hour on construction paper and use off-the-shelf electronics. The line between consumer appliances and robotics is already blurring with products like the Roomba vacuum cleaner. Some of these ubiquitous robots will be used in search-and-rescue operations while others might help us fold the laundry.
Professor Julie Shah, who leads CSAIL’s Interactive Robotics Group, talked about her research on Factories of the Near Future, where humans and robots will be able to work in close proximity to each other. Such human-robot interactions make it possible to better harness the relative strengths of each, and will help introduce robot assistants into many previously human-only domains.
In a recent paper, Toward Safe Close-Proximity Human-Robot Interaction with Standard Industrial Robots, Shah and her co-authors addressed the safety issues involved in close proximity human-robot interactions. Their research includes the development of “a real-time safety system capable of allowing safe human-robot interaction at very low distances of separation, without the need for robot hardware modification or replacement.”
She also talked about the issues involved in human-robot cross training, the subject of another recent paper. “When humans work in teams, it is crucial for the members to develop fluent team behavior. We believe that the same holds for robot teammates, if they are to perform in a similarly fluent manner as members of a human-robot team.” Shah’s group is adapting methods that have proved effective in training human teams to now train human-robot teams.
The last presentation I’d like to discuss is by Gill Pratt, Program Manager of the DARPA Robotics Challenge (DRC). The DARPA Grand Challenge, a prize competition introduced a decade ago, helped accelerate the development of driverless cars. The DARPA Robotics Challenge aims to have a similar impact on the development of “ground robots capable of executing complex tasks in dangerous, degraded, human-engineered environments. Competitors in the DRC are expected to focus on robots that can use standard tools and equipment commonly available in human environments, ranging from hand tools to vehicles, with an emphasis on adaptability to tools with diverse specifications.”
The DRC is another example of human-robot collaboration. In its supervised autonomy environment, human operators guide the robot in completing its various tasks using relatively low fidelity communications that mimic the degraded situation in natural or man-made disasters. Pratt’s presentation included impressive videos of the eight finalist teams in action. In this recent video he explained the goals of the DRC:
“DARPA’s role is to spur innovation. And we do it by focused, short term efforts. We pick things that are not impossible, but also not very low risk. We take very high risk gambles, and those risks have tremendous payoffs. . . My expectation is that the robots are going to be slow. What we’re looking for right now is for the teams to just do as well as roughly a one year old child. If we can get there, then we think that we have good reason to believe that some of these teams with continued persistence for another year will actually be able to demonstrate robots that show the utility that these things might have in a real disaster scenario.”
The conference was truly thought provoking. Let me conclude with my overall impressions of what I heard.
As Brynjolfsson said, advances in AI software, robotics and related technologies are quite amazing. The conference gave us a glimpse of the leading-edge research that will lead to future such advances. But, their successful deployment in the marketplace, - in production and at scale, - will be even more challenging.
I’m pretty sure that a number of the innovations in Rus’ Distributed Robotics Lab will find their way to the marketplace, - as increasingly smart appliances, wearable devices, tools and apps. I also really liked the human-robot collaborations discussed by Shah and Pratt. Their pragmatic projects view robots as highly sophisticated tools that help and complement humans, each performing those tasks they are best at.
What about the prospects for fully autonomous machines, such as self-driving cars? Some feel that they will be all around us within a decade or so, navigating our present roads right along human-driven cars. I’m not so sure. I think that self-driving cars are in a kind-of hype cycle. Lots and lots of issues, most of them not technical, remain to be worked out before a self-driving car will take you to work on regular city streets while you’re reading the paper.
The research is very important. The technologies being developed for self-driving cars will significantly improve the safety of our cars, as is already happening with collision avoidance systems, lane departure warning systems, and related innovations. In all these cases, the human drivers are in control and would be responsible if an accident occurs, as they would be today while using cruise control.
I can also envision self-driving cars in a more controlled environment, such as instrumented lanes in cities or highways where especially designed and licensed vehicles can be driven in a kind of autopilot mode. However, as is the case with airline pilots, the vehicle drivers would continue to be in charge and are expected to assume manual control as needed.
At some point in the future, society might be comfortable with truly self-driven vehicles and similar autonomous machines. I truly don’t know when that might be. Such times will arrive when the no-longer-so-amazing technologies and our human societies have jointly co-evolved and adapted to each other, as was the case with electricity and cars in the past, and with the Internet and smartphones more recently. As Marshal McLuhan observed in the 1960s: We shape our tools and they in turn shape us.