A Tech-Fueled Approach to Collective Problem-SolvingDirector of the Governance Lab and Professor of Engineering at New York University
There are more people learning foreign languages on the Duolingo platform than there are people learning foreign languages in the whole U.S. public school system. Designed by Carnegie Mellon professor Luis von Ahn, DuoLingo has reduced the time it takes to achieve basic mastery of a foreign language to 34 hours, earning it 200 million active users learning 19 languages. Why is it so successful?
Duolingo combines human and machine intelligence with powerful results. As Mr. Von Ahn explained in an interview, “if we want to know whether we should teach plurals before adjectives, for the next 50,000 users, we’ll teach half of them plurals before adjectives, half the other way around and then we measure which of these groups learns better. We figure out in a couple of days which is more effective and then switch everyone to that.”
With learners undertaking 6 billion exercises every month, Duolingo connects people and computing power into a system that works better than either machines or humans working alone. It is a paradigmatic example of what is called collective intelligence.
While classically, collective intelligence is how people think and act in groups, today collective intelligence includes how new technologies, especially the Internet, can help us become smarter together than we are alone and mobilize human intelligence at scale.
Whereas once the best examples of collective intelligence involved collections of people coming together in physical space, today the Internet makes it possible to combine humans and machines, organizations and networks, across a distance.
The classic example is Wikipedia, a collective intelligence platform to organize human knowledge. There is also WikiHow, which organizes hundreds of thousands of how-to videos created by volunteers in 17 languages. YouTube benefits from 300 hours of user-generated video uploaded every minute. Mechanical Turk is Amazon’s platform for organizing the labor of 500,000 people, undertaking and getting paid for small tasks posted by hundreds of requesting employers each day, making Amazon’s Mechanical Turk one of the largest outsourcing platforms in the United States.
The value of the Internet for connecting people to aggregate large quantities of distributed information is obvious. But many examples of collective intelligence are less intentionally collaborative. Collective intelligence also includes networking large numbers of people to aggregate their individual experience, expertise or actions without their intending to collaborate.
Making Sense of Dispersed Knowledge
Google Maps and its companion traffic app Waze depend on information gathered anonymously and passively from cell phone users. Prediction markets, such as Iowa Electronic Markets, aggregate the judgments of their distributed participants without the need for the individuals to work together. The platform efficiency makes sense of their dispersed knowledge, usually resulting in guesses about stock prices, election results, outcomes of world events or size of box office draws of new movies more accurately than the predictions of either pundits or opinion polls.
As Geoff Mulgan writes in Big Mind, his excellent 2017 book on collective intelligence, it is an anachronism to assume that “intelligence resides primarily in the space inside the human skull.”
Despite the promise of collective intelligence, the reality is that most companies and governments still make decisions top-down and behind closed doors.
For this reason, companies and governments alike are using the practices of crowdsourcing to try to structure collective intelligence to their benefit, especially as research shows that when average participants are asked to perform tasks with specific instructions, their performance is equal to or better than the performance of experts.
Although the website Patients Like Me aggregates the experiences of those diagnosed with a disease for mutual benefit and the Internet Movie Database collects film information from millions of movie buffs, not every use of the crowd makes us smarter or better off.
Top-Down Is Hard To Shake
Despite the promise of collective intelligence, the reality is that most companies and governments still make decisions top-down and behind closed doors, relying on politicians and bureaucrats to tackle our most pressing problems.
Public officials are often ill-equipped to know how to regulate the complex issues of our time, especially those involving scientific advances, such as climate change or autonomous vehicles, which raise myriad ethical, moral, political, legal, regulatory and social questions.
From Crowdsourcing to CrowdLaw
But what if the technologies of collective intelligence could enable more individuals—not only interest groups—to inform legislative and policymaking processes? What if we knew how to mobilize the crowd to make law and policy? What if we could use the “crowdlaw“ collective intelligence efforts to make our smart cities even smarter and give each of us the ability to participate in governing?
There are some exciting examples of crowdlaw on the horizon of public institutions turning to collective intelligence to govern differently. In the Republic of China, the vTaiwan experimental e-consultation platform enables the broader public to participate in an ongoing process of problem identification. So far, 26 national issues, including the regulation of telemedicine, online education, telework, company law and Uber have been discussed with over 200,000 people participating—a small but auspicious start.
In Iceland, Better Reykjavik is a crowdlaw web platform for “idea generation” and “policy crowdsourcing” for citizens to present and discuss ideas related to the services and operations of the city of Reykjavik. The website has been used by 20 percent of Iceland’s population, and over half of those registered on the site use it regularly.
The Internet also offers a way to engage the public in evaluating what’s working. In 2016, Brazil’s comptroller launched an experimental project called the Projeto Controladoria na Escola to engage students in 10 public schools in Brasilia in the process of auditing school infrastructure, mapping commonly raised issues and fostering civic education in schools. In one school alone, the students identified 115 issues, and within just three months, 45 percent of the issues were fixed either by the department of education or by the students and school management themselves.
A Better Way To Run an Organization—and a Country?
Over the next fifty years, we will face challenges greater than our predecessors, and we will need to run our institutions—be they corporations or governments—differently if we are to be able to respond. People have the passion and the know-how to participate. The platforms make it possible. Now all we need are the institutions that know how and want to make use of our collective intelligence for problem-solving.