Podcast – The Power of Collective Intelligence – In conversation with Dr Emile Servan-Schreiber

“Diversity” is the buzzword of the moment, but it’s being treated like a moral imperative, rather than something that can help businesses become better and smarter. France’s Dr Emile Servan-Schreiber, the managing director of Hypermind, says the moral position is only part of the story. As he outlined in his book Supercollectif, diverse groups are measurably smarter and able to tackle more complex problems than homogenous ones.

What is collective intelligence and why is it so importance to unlock it in our organisations?

There are so many different answers to that. One is that collective intelligence is the reason we exist. If you think about it, a single human being would not be able to survive in ancient times against the fiercer carnivorous animals with longer nails and bigger teeth that could outrun us. The only reason we still exist is because we were able to be intelligent as a group. We could stand together as a group and hunt big game that would be able to feed us. You take the smartest person, put him in the savannah and he wouldn’t survive for 24 hours. You think of history as being made by genius adventurers, individual men and women, but in fact it’s done by everybody together.

How are we able collectively to solve problems? That’s collective intelligence. Historically, it was discovered as a phenomenon by looking at insect colonies. That’s the first obvious example of collective intelligence. A small ant or termite doesn’t do a lot individually, but as a colony there is an emergent property. Collectively, they are intelligent.

Dr Emile Servan-Schreiber and his latest book, soon available in English

What’s new is that now we can measure the power of collective intelligence.

Since the internet arrived on the scene, we are able to connect humans on a massive scale and see the same kind of emergent intelligence. We were able to see it in insects, but now we can see it in humans and study it more easily. Scientists thought about measuring the intelligence of human groups only very recently and that’s just about ten years ago.

MIT and Carnegie Mellon decided to take groups of people and have them take IQ tests – which were invented in France 100 years ago – and when you do that, you realise that you can look at the group’s ability to solve a problem and how long it takes to solve the problem. Then you put a number on that and call it IQ. The interesting thing about this number is that it’s predictive of the group’s ability to solve the next problem, just like your IQ is. Not only do groups have intelligence but, just like individuals, different groups have different IQs. Some of them are smarter than others and that’s when it becomes really interesting. How do you make a group smarter?

What makes one group smarter than another is not how many smart people there are in the group. And it’s not that you have a super smart person in the group, although that’s better than having many smart people for some reason, which is a little counter intuitive. But in fact what characterises the smartest groups is that they have more women in them, which is a little bit surprising – with all due respect. The scientists were a hybrid group of men and women but they were still a little bit surprised by this result. They decided to dig further to find out what’s behind this strange result, which is very obvious in the data. If your group has more than 50% women, it has a higher than average IQ. Groups that have less than 50% have lower IQs compared to the others. What’s behind that is pretty simple.

If your group has more than 50% women, it has a higher than average IQ. Groups that have less than 50% have lower IQs compared to the others.

Dr Emile Servan-Schreiber

In groups where there are more women, there is a better distribution of speaking time. The more women you have, the fewer men try to monopolise speech in the group. The other side is that women have, in general, a better ability to listen. In groups with more women, more people can express themselves and more people listen when they do, and that’s what makes the group smart. Everybody gets to take part.

Is our intelligence fundamentally different, depending on our cultural backgrounds?

Yes, obviously culture has a huge impact on how you think about things. We could see that in the pandemic as well; Asians think about the collective and what is the good behaviour of the citizen very differently from a French person or an American.

When you’re trying to get at the truth of something, it is very important to get a group of people who think differently about the problem. If the problem is really a problem, it’s complicated, so no one has the one best answer. If you can assemble a group of people who can look at the problem through different values and can share what each of them view around the problem, then as a group they are going to be much smarter.

Diversity gets a lot of good publicity these days, where everybody says we should be more diverse. People talk about it as some kind of moral imperative; that because there are 50% of women in the world, there should be 50% of women on boards.

In fact, the power of diversity comes not from it being a moral imperative but a law of nature.

Dr Emile Servan-Schreiber

You can actually express it as a mathematical formula, in something called the ‘diversity prediction theorem’. It’s quite simple algebra and it says the error the group will make when estimating something will depend on two things – it will depend equally on those two things. Firstly, it depends on the errors that the individuals in the group are going to make. The group is going to be smarter if the people in the group are a bit smarter as well. The other element is that the group is even smarter if the opinions of the people in the group diverge. The less they agree, the more likely it is the group will be smarter, because it takes account of more different opinions, each of which has a bit of truth and each of which contains mistakes, biases and subjective errors that people make.

It’s really important when you assemble a group to have people who know something about the situation, but who know different things about the situation. It’s really important to encourage people to really say what they think, rather than to be in consensus with other people. Consensus kills intelligence.

We all make different mistakes and those mistakes correct each other.

The discovery of the “wisdom of crowds”.

We invented IQ in France in 1905. In 1906 was the first discovery of what we now call the “wisdom of crowds”, or the intelligence of the group on a massive scale. It was discovered by Francis Galton, who was a cousin of Charles Darwin. He was a genius statistician, one of the best of the time. At the end of his life he made a final discovery, when he happened on a country fair where people were trying to guess the weight of an ox.

Galton’s data: the bigger the crowd, the better the predictions. The average error rate for an individual is 4.5%. Ask 10 people and the now collective average error rate is divided by three. Big crowds lead to a close to zero error rate.

Galton came on this crowd of hundreds of people each one paying a few shillings to write his best estimate on a piece of paper and Galton said, “that’s really interesting. We’re discussing promoting more democracy in England”. He was a member of the elite at the time so he was a bit sceptical of more democracy and he said, “well that’s a wonderful occasion to test whether the voice of the people has anything to say, because those people trying to guess the weight of the ox don’t know more about ox than the average voter knows about politics”. He got the tickets home and put them in order from the lowest to the highest, and said, the median estimate, that’s the voice of the people. Then he realised that this number was almost right on the actual weight of the ox. An error of 0.05% or something ridiculous and he said, “why not democracy?”.

Anybody who has spent time on Twitter can also wonder about people getting together to be more stupid than more clever. How can we make a group more intelligent than its worst individuals?

The recipe to make the group smart is to assemble people who are likely to think differently. Secondly, there is no point in people thinking differently if they don’t express it. So you need a process that encourages people to express what they think – especially if it doesn’t conform to what other people think. The third thing is once everybody has expressed their opinion, you need to make a decision in a way that aggregates all the information that’s been put on the table. It could be an election, it could be taking the average of taking various estimates, or it could be that people bet on the outcome.

In your book you say we also need diversity because we pay more attention to people who are less like us.

We distrust people who don’t look like us. If you’re around people who are just like yourself, your natural tendency is to stop thinking and to let other people think. But if you’re among people who are different – black, white, men, women, French, English, whatever, as long as there is some difference that is visible – then you won’t feel that you are accurately represented by others. You become more responsible for expressing your opinion and you do more independent thinking – which is the key to making a group smart.

Dr Emile Servan-Schreiber holds a PhD in cognitive psychology from Carnegie-Mellon University in the USA, as well as undergraduate degrees in mathematics and computing. He is the managing director of Hypermind which operates in both France and the USA, which creates prediction markets.

This transcript has been edited and condensed and only represents part of a fascinating conversation. Listen to the entire conversation here:

The ARENI Global In Conversation series showcases some of the world’s most interesting speakers and issues.

Our research, publications and events are only possible thanks to people like you. If you have the capacity to do so, please consider becoming a member.