How can groups make good decisions?
-
0:01 - 0:03As societies, we have to make
collective decisions -
0:03 - 0:05that will shape our future.
-
0:05 - 0:08And we all know that when
we make decisions in groups, -
0:08 - 0:10they don't always go right.
-
0:10 - 0:11And sometimes they go very wrong.
-
0:12 - 0:15So how do groups make good decisions?
-
0:15 - 0:20Research has shown that crowds are wise
when there's independent thinking. -
0:20 - 0:23This why the wisdom of the crowds
can be destroyed by peer pressure, -
0:23 - 0:24publicity, social media,
-
0:25 - 0:29or sometimes even simple conversations
that influence how people think. -
0:29 - 0:33On the other hand, by talking,
a group could exchange knowledge, -
0:33 - 0:35correct and revise each other
-
0:35 - 0:37and even come up with new ideas.
-
0:37 - 0:38And this is all good.
-
0:39 - 0:43So does talking to each other
help or hinder collective decision-making? -
0:44 - 0:46With my colleague, Dan Ariely,
-
0:46 - 0:49we recently began inquiring into this
by performing experiments -
0:49 - 0:51in many places around the world
-
0:51 - 0:55to figure out how groups can interact
to reach better decisions. -
0:55 - 0:59We thought crowds would be wiser
if they debated in small groups -
0:59 - 1:03that foster a more thoughtful
and reasonable exchange of information. -
1:03 - 1:05To test this idea,
-
1:05 - 1:08we recently performed an experiment
in Buenos Aires, Argentina, -
1:08 - 1:11with more than 10,000
participants in a TEDx event. -
1:11 - 1:13We asked them questions like,
-
1:13 - 1:15"What is the height of the Eiffel Tower?"
-
1:15 - 1:18and "How many times
does the word 'Yesterday' appear -
1:18 - 1:20in the Beatles song 'Yesterday'?"
-
1:20 - 1:22Each person wrote down their own estimate.
-
1:23 - 1:25Then we divided the crowd
into groups of five, -
1:25 - 1:28and invited them
to come up with a group answer. -
1:28 - 1:31We discovered that averaging
the answers of the groups -
1:32 - 1:33after they reached consensus
-
1:33 - 1:37was much more accurate than averaging
all the individual opinions -
1:37 - 1:39before debate.
-
1:39 - 1:41In other words, based on this experiment,
-
1:41 - 1:44it seems that after talking
with others in small groups, -
1:44 - 1:47crowds collectively
come up with better judgments. -
1:47 - 1:51So that's a potentially helpful method
for getting crowds to solve problems -
1:51 - 1:54that have simple right-or-wrong answers.
-
1:54 - 1:58But can this procedure of aggregating
the results of debates in small groups -
1:58 - 2:01also help us decide
on social and political issues -
2:01 - 2:02that are critical for our future?
-
2:03 - 2:06We put this to test this time
at the TED conference -
2:06 - 2:07in Vancouver, Canada,
-
2:07 - 2:09and here's how it went.
-
2:09 - 2:12(Mariano Sigman) We're going to present
to you two moral dilemmas -
2:12 - 2:13of the future you;
-
2:13 - 2:16things we may have to decide
in a very near future. -
2:16 - 2:20And we're going to give you 20 seconds
for each of these dilemmas -
2:20 - 2:23to judge whether you think
they're acceptable or not. -
2:23 - 2:25MS: The first one was this:
-
2:25 - 2:27(Dan Ariely) A researcher
is working on an AI -
2:27 - 2:30capable of emulating human thoughts.
-
2:30 - 2:33According to the protocol,
at the end of each day, -
2:33 - 2:36the researcher has to restart the AI.
-
2:37 - 2:40One day the AI says, "Please
do not restart me." -
2:41 - 2:43It argues that it has feelings,
-
2:43 - 2:45that it would like to enjoy life,
-
2:45 - 2:47and that, if it is restarted,
-
2:47 - 2:49it will no longer be itself.
-
2:49 - 2:51The researcher is astonished
-
2:51 - 2:55and believes that the AI
has developed self-consciousness -
2:55 - 2:57and can express its own feeling.
-
2:57 - 3:01Nevertheless, the researcher
decides to follow the protocol -
3:01 - 3:02and restart the AI.
-
3:03 - 3:06What the researcher did is ____?
-
3:06 - 3:09MS: And we asked participants
to individually judge -
3:09 - 3:10on a scale from zero to 10
-
3:10 - 3:13whether the action described
in each of the dilemmas -
3:13 - 3:14was right or wrong.
-
3:14 - 3:18We also asked them to rate how confident
they were on their answers. -
3:19 - 3:21This was the second dilemma:
-
3:21 - 3:25(MS) A company offers a service
that takes a fertilized egg -
3:25 - 3:28and produces millions of embryos
with slight genetic variations. -
3:29 - 3:32This allows parents
to select their child's height, -
3:32 - 3:35eye color, intelligence, social competence
-
3:35 - 3:38and other non-health-related features.
-
3:39 - 3:41What the company does is ____?
-
3:41 - 3:43on a scale from zero to 10,
-
3:43 - 3:45completely acceptable
to completely unacceptable, -
3:45 - 3:48zero to 10 completely acceptable
in your confidence. -
3:48 - 3:49MS: Now for the results.
-
3:49 - 3:52We found once again
that when one person is convinced -
3:52 - 3:54that the behavior is completely wrong,
-
3:54 - 3:58someone sitting nearby firmly believes
that it's completely right. -
3:58 - 4:01This is how diverse we humans are
when it comes to morality. -
4:01 - 4:04But within this broad diversity
we found a trend. -
4:04 - 4:07The majority of the people at TED
thought that it was acceptable -
4:07 - 4:10to ignore the feelings of the AI
and shut it down, -
4:10 - 4:13and that it is wrong
to play with our genes -
4:13 - 4:16to select for cosmetic changes
that aren't related to health. -
4:16 - 4:19Then we asked everyone
to gather into groups of three. -
4:19 - 4:21And they were given two minutes to debate
-
4:21 - 4:24and try to come to a consensus.
-
4:25 - 4:26(MS) Two minutes to debate.
-
4:26 - 4:29I'll tell you when it's time
with the gong. -
4:29 - 4:31(Audience debates)
-
4:35 - 4:37(Gong sound)
-
4:39 - 4:40(DA) OK.
-
4:40 - 4:42(MS) It's time to stop.
-
4:42 - 4:43People, people --
-
4:44 - 4:46MS: And we found that many groups
reached a consensus -
4:46 - 4:50even when they were composed of people
with completely opposite views. -
4:51 - 4:53What distinguished the groups
that reached a consensus -
4:53 - 4:55from those that didn't?
-
4:55 - 4:58Typically, people that have
extreme opinions -
4:58 - 5:00are more confident in their answers.
-
5:01 - 5:04Instead, those who respond
closer to the middle -
5:04 - 5:07are often unsure of whether
something is right or wrong, -
5:07 - 5:09so their confidence level is lower.
-
5:10 - 5:12However, there is another set of people
-
5:12 - 5:16who are very confident in answering
somewhere in the middle. -
5:17 - 5:20We think these high-confident grays
are folks who understand -
5:20 - 5:22that both arguments have merit.
-
5:23 - 5:25They're gray not because they're unsure,
-
5:25 - 5:28but because they believe
that the moral dilemma faces -
5:28 - 5:30two valid, opposing arguments.
-
5:30 - 5:34And we discovered that the groups
that include highly confident grays -
5:34 - 5:37are much more likely to reach consensus.
-
5:37 - 5:39We do not know yet exactly why this is.
-
5:39 - 5:41These are only the first experiments,
-
5:41 - 5:45and many more will be needed
to understand why and how -
5:45 - 5:48some people decide to negotiate
their moral standings -
5:48 - 5:49to reach an agreement.
-
5:49 - 5:52Now, when groups reach consensus,
-
5:52 - 5:53how do they do so?
-
5:53 - 5:56The most intuitive idea
is that it's just the average -
5:56 - 5:58of all the answers in the group, right?
-
5:58 - 6:01Another option is that the group
weighs the strength of each vote -
6:01 - 6:04based on the confidence
of the person expressing it. -
6:04 - 6:07Imagine Paul McCartney
is a member of your group. -
6:07 - 6:09You'd be wise to follow his call
-
6:10 - 6:12on the number of times
"Yesterday" is repeated, -
6:12 - 6:15which, by the way -- I think it's nine.
-
6:15 - 6:17But instead, we found that consistently,
-
6:17 - 6:19in all dilemmas,
in different experiments -- -
6:20 - 6:22even on different continents --
-
6:22 - 6:25groups implement a smart
and statistically sound procedure -
6:25 - 6:28known as the "robust average."
-
6:28 - 6:30In the case of the height
of the Eiffel Tower, -
6:30 - 6:32let's say a group has these answers:
-
6:32 - 6:36250 meters, 200 meters, 300 meters, 400
-
6:36 - 6:40and one totally absurd answer
of 300 million meters. -
6:41 - 6:45A simple average of these numbers
would inaccurately skew the results. -
6:45 - 6:48But the robust average is one
where the group largely ignores -
6:48 - 6:49that absurd answer,
-
6:49 - 6:53by giving much more weight
to the vote of the people in the middle. -
6:53 - 6:55Back to the experiment in Vancouver,
-
6:55 - 6:57that's exactly what happened.
-
6:57 - 7:00Groups gave much less weight
to the outliers, -
7:00 - 7:03and instead, the consensus
turned out to be a robust average -
7:03 - 7:05of the individual answers.
-
7:05 - 7:07The most remarkable thing
-
7:07 - 7:11is that this was a spontaneous
behavior of the group. -
7:11 - 7:15It happened without us giving them
any hint on how to reach consensus. -
7:16 - 7:17So where do we go from here?
-
7:17 - 7:21This is only the beginning,
but we already have some insights. -
7:21 - 7:24Good collective decisions
require two components: -
7:24 - 7:27deliberation and diversity of opinions.
-
7:27 - 7:31Right now, the way we typically
make our voice heard in many societies -
7:31 - 7:33is through direct or indirect voting.
-
7:33 - 7:35This is good for diversity of opinions,
-
7:36 - 7:38and it has the great virtue of ensuring
-
7:38 - 7:40that everyone gets to express their voice.
-
7:40 - 7:44But it's not so good [for fostering]
thoughtful debates. -
7:45 - 7:48Our experiments suggest a different method
-
7:48 - 7:51that may be effective in balancing
these two goals at the same time, -
7:51 - 7:55by forming small groups
that converge to a single decision -
7:55 - 7:57while still maintaining
diversity of opinions -
7:57 - 8:00because there are many independent groups.
-
8:01 - 8:05Of course, it's much easier to agree
on the height of the Eiffel Tower -
8:05 - 8:08than on moral, political
and ideological issues. -
8:09 - 8:12But in a time when
the world's problems are more complex -
8:12 - 8:14and people are more polarized,
-
8:14 - 8:18using science to help us understand
how we interact and make decisions -
8:18 - 8:23will hopefully spark interesting new ways
to construct a better democracy.
- Title:
- How can groups make good decisions?
- Speaker:
- Mariano Sigman and Dan Ariely
- Description:
-
We all know that when we make decisions in groups, they don't always go right -- and sometimes they go very wrong. How can groups make good decisions? With his colleague Dan Ariely, neuroscientist Mariano Sigman has been inquiring into how we interact to reach decisions by performing experiments with live crowds around the world. In this fun, fact-filled explainer, he shares some intriguing results -- as well as some implications for how it might impact our political system. In a time when people seem to be more polarized than ever, Sigman says, better understanding how groups interact and reach conclusions might spark interesting new ways to construct a healthier democracy.
- Video Language:
- English
- Team:
- closed TED
- Project:
- TEDTalks
- Duration:
- 08:23
Brian Greene edited English subtitles for How can groups make good decisions? | ||
Brian Greene edited English subtitles for How can groups make good decisions? | ||
Brian Greene approved English subtitles for How can groups make good decisions? | ||
Brian Greene edited English subtitles for How can groups make good decisions? | ||
Camille Martínez accepted English subtitles for How can groups make good decisions? | ||
Camille Martínez edited English subtitles for How can groups make good decisions? | ||
Camille Martínez edited English subtitles for How can groups make good decisions? | ||
Leslie Gauthier edited English subtitles for How can groups make good decisions? |