Harnessing the Collective Intelligence of Humans and Machines
For over a decade, MIT professor Thomas Malone has been conducting pioneering research on the collective intelligence of groups. Malone is the founding director of the MIT Center for Collective Intelligence, and has authored and co-authored a number of seminal articles on the subject. A few weeks ago, Malone published Superminds: The Surprising Power of People and Computers Thinking Together, - a book that nicely explains the concept of collective intelligence and how our increasingly capable machines are now complementing the intelligence of groups of people. The book illustrates these concepts with a wide variety of examples and case studies.
He defines a supermind as “a group of individuals acting in ways that seem intelligent.” A supermind has measurable properties just as individual do. It can have a life of its own beyond those of its individuals constituents. And, as Malone points out, throughout history, groups of people working together as superminds have been responsible for the vast majority of human achievements.
Intelligence plays a central role in Superminds, so the book’s first chapter discusses various views of the term. For example, in 1994 the Wall Street Journal published the Mainstream View of Intelligence, an article that reflected the consensus of 52 leading academic researchers in fields associated with intelligence, which included this definition:
“Intelligence is a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings - ‘catching on,’ ‘making sense’ of things, or ‘figuring out’ what to do.”
This is a very good definition of general intelligence, - the ability to effectively address a wide range of goals in different environments. It’s the kind of intelligence that’s long been measured in IQ tests, and that, for the foreseeable future, only humans have.
On the other hand, specialized intelligence, - the ability to effectively address well-defined, specific goals in a given environment, - is the kind of task-oriented intelligence that’s part of many human jobs. In the past several years, our increasingly capable machines have become quite proficient at handling a variety of such specialized intelligent tasks.
The book raises a number of important questions, from the more practical, - how can humans and machines work together in new ways?, - to the more philosophical, - can superminds exhibit something like consciousness? I’d like to focus my brief discussions below on two of the book’s key questions: can you objectively measure the intelligence of a group or supermind?; and, how can computers make superminds smarter?
Can you measure the IQ of a group?
Early in the 20th century it was discovered that people’s cognitive abilities tend to be similar across a wide variety of tasks. That is, if you perform well in one such cognitive task you are likely to also perform well in others even though they may be quite different. Such a general cognitive intelligence can be statistically captured in something called the g factor, which is a major component of tests measuring the intelligence quotient (IQ) of individuals.
Malone and his collaborators wondered if a similar definition of intelligence could apply to groups of people. That is, do groups, like individuals, exhibit characteristic levels of intelligence which can be measured and used to predict the group’s performance across a wide variety of cognitive tasks. And if so, can they find a statistically significant factor which measures the group’s collective intelligence analogous to an individual’s IQ.
To answer these questions they conducted two separate research studies. In their initial set of studies, they randomly assigned nearly 700 volunteers into groups of two to five members. Each group worked together on a diverse set of short tasks selected to represent the kinds of problems that groups work on in the real world. These included tasks requiring logical analysis such as solving visual puzzles; tasks emphasizing collective brainstorming and moral judgements; and tasks based on coordination and planning such as negotiating over limited resources.
The study looked at group cohesion, motivation and satisfaction, but none of them worked. It measured the individual IQs of each of the participants, and found that the average and maximum intelligence of group members was correlated with the group’s collective intelligence, but only moderately so. Just having a bunch of smart people in a group wasn’t quite enough to make the group smart.
The study did find three statistically significant factors that predicted how well each group would do on a wide range of tasks. The key factor was the average social sensitivity of group members, that is, the ability to read each others emotional states as measured by the Reading the Mind in the Eyes test originally developed at the Autism Research Center in the University of Cambridge. It also found that groups in which a few people dominated the conversation did not perform as well as those groups where speaking and contributions were more evenly distributed. Finally, the study found that collective intelligence positively correlated with the proportion of women in the group. This is likely because as previous research has shown, women generally score higher than men in the Reading-the-Mind social sensitivity tests.
In a later study, the research team replicated its earlier findings, but with a twist. They wanted to explore whether groups that worked online instead of face-to-face also exhibited collective intelligence. To do so, they assembled 68 teams, half of which worked face to face like those in their earlier studies, and half worked online with no ability to see the other group members. They found that whether online or off, some teams consistently outperformed the others. And, just like in the earlier studies, the best performing teams exhibited higher social perceptiveness skills.
How can computers make superminds smarter?
In a 2012 online conversation, Malone discussed the fundamental research challenge that led him to found the MIT Center for Collective Intelligence: “How can people and computers be connected so that - collectively - they act more intelligently than any individuals, groups, or computers have ever done before?” Not surprisingly, this big question is the key theme of Superminds, especially given the dramatic advances in AI in the intervening years.
Our increasingly smart machines are now being successfully applied to tasks that not long ago were viewed as the exclusive domain of humans, - e.g., machine translation, natural language processing, autonomous vehicles. Their deep, specialized intelligence enables them to effectively address a variety of highly sophisticated tasks. It’s quite possible that sometime in the future, there will be AI machines with human-like general intelligence that are as smart and broadly adaptable as people. But, no one can tell when, or even if, that time will come.
In the meantime, we’ll need to use AI in combination with people. Humans can supply the general intelligence and whatever other skills machines don’t have, and machines can supply the vast information, computational power and other specialized capabilities that people don’t have.
“For the foreseeable future, therefore, there is another way of using IT that will be even more important than just creating better AI: creating groups of people and computers that, together, are far more collectively intelligent than was ever possible before,” writes Malone. “While we often overestimate the potential of AI in doing this, I think we often underestimate the potential power of hyperconnectivity among the seven billion or so amazingly powerful information processors called human brains that are already on our planet, not to mention the millions of other computers that don’t include AI… As all the people and computers on our planet get more and more closely connected, it's becoming increasingly useful to think of all the people and computers on the planet as a kind of global brain.”