[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:00.55,0:00:02.100,Default,,0000,0000,0000,,As societies, we have to make\Ncollective decisions
Dialogue: 0,0:00:03.02,0:00:04.59,Default,,0000,0000,0000,,that will shape our future.
Dialogue: 0,0:00:05.09,0:00:07.84,Default,,0000,0000,0000,,And we all know that when\Nwe make decisions in groups,
Dialogue: 0,0:00:07.87,0:00:09.51,Default,,0000,0000,0000,,they don't always go right.
Dialogue: 0,0:00:09.53,0:00:11.49,Default,,0000,0000,0000,,And sometimes they go very wrong.
Dialogue: 0,0:00:12.32,0:00:14.74,Default,,0000,0000,0000,,So how do groups make good decisions?
Dialogue: 0,0:00:15.23,0:00:19.56,Default,,0000,0000,0000,,Research has shown that crowds are wise\Nwhen there's independent thinking.
Dialogue: 0,0:00:19.58,0:00:22.78,Default,,0000,0000,0000,,This why the wisdom of the crowds\Ncan be destroyed by peer pressure,
Dialogue: 0,0:00:22.81,0:00:24.50,Default,,0000,0000,0000,,publicity, social media,
Dialogue: 0,0:00:24.52,0:00:28.56,Default,,0000,0000,0000,,or sometimes even simple conversations\Nthat influence how people think.
Dialogue: 0,0:00:29.06,0:00:33.02,Default,,0000,0000,0000,,On the other hand, by talking,\Na group could exchange knowledge,
Dialogue: 0,0:00:33.04,0:00:34.82,Default,,0000,0000,0000,,correct and revise each other
Dialogue: 0,0:00:34.85,0:00:36.64,Default,,0000,0000,0000,,and even come up with new ideas.
Dialogue: 0,0:00:36.66,0:00:37.96,Default,,0000,0000,0000,,And this is all good.
Dialogue: 0,0:00:38.50,0:00:43.17,Default,,0000,0000,0000,,So does talking to each other\Nhelp or hinder collective decision-making?
Dialogue: 0,0:00:43.75,0:00:45.54,Default,,0000,0000,0000,,With my colleague, Dan Ariely,
Dialogue: 0,0:00:45.57,0:00:49.14,Default,,0000,0000,0000,,we recently began inquiring into this\Nby performing experiments
Dialogue: 0,0:00:49.16,0:00:50.94,Default,,0000,0000,0000,,in many places around the world
Dialogue: 0,0:00:50.97,0:00:55.24,Default,,0000,0000,0000,,to figure out how groups can interact\Nto reach better decisions.
Dialogue: 0,0:00:55.26,0:00:58.81,Default,,0000,0000,0000,,We thought crowds would be wiser\Nif they debated in small groups
Dialogue: 0,0:00:58.84,0:01:02.76,Default,,0000,0000,0000,,that foster a more thoughtful\Nand reasonable exchange of information.
Dialogue: 0,0:01:03.39,0:01:04.59,Default,,0000,0000,0000,,To test this idea,
Dialogue: 0,0:01:04.62,0:01:07.86,Default,,0000,0000,0000,,we recently performed an experiment\Nin Buenos Aires, Argentina,
Dialogue: 0,0:01:07.89,0:01:10.89,Default,,0000,0000,0000,,with more than 10,000\Nparticipants in a TEDx event.
Dialogue: 0,0:01:11.49,0:01:12.95,Default,,0000,0000,0000,,We asked them questions like,
Dialogue: 0,0:01:12.97,0:01:14.92,Default,,0000,0000,0000,,"What is the height of the Eiffel Tower?"
Dialogue: 0,0:01:14.95,0:01:17.68,Default,,0000,0000,0000,,and "How many times\Ndoes the word 'Yesterday' appear
Dialogue: 0,0:01:17.70,0:01:20.00,Default,,0000,0000,0000,,in the Beatles song 'Yesterday'?"
Dialogue: 0,0:01:20.02,0:01:22.32,Default,,0000,0000,0000,,Each person wrote down their own estimate.
Dialogue: 0,0:01:22.77,0:01:25.27,Default,,0000,0000,0000,,Then we divided the crowd\Ninto groups of five,
Dialogue: 0,0:01:25.29,0:01:28.02,Default,,0000,0000,0000,,and invited them\Nto come up with a group answer.
Dialogue: 0,0:01:28.50,0:01:31.49,Default,,0000,0000,0000,,We discovered that averaging\Nthe answers of the groups
Dialogue: 0,0:01:31.52,0:01:33.07,Default,,0000,0000,0000,,after they reached consensus
Dialogue: 0,0:01:33.09,0:01:37.33,Default,,0000,0000,0000,,was much more accurate than averaging\Nall the individual opinions
Dialogue: 0,0:01:37.35,0:01:38.52,Default,,0000,0000,0000,,before debate.
Dialogue: 0,0:01:38.55,0:01:41.18,Default,,0000,0000,0000,,In other words, based on this experiment,
Dialogue: 0,0:01:41.20,0:01:44.34,Default,,0000,0000,0000,,it seems that after talking\Nwith others in small groups,
Dialogue: 0,0:01:44.36,0:01:47.07,Default,,0000,0000,0000,,crowds collectively\Ncome up with better judgments.
Dialogue: 0,0:01:47.09,0:01:50.62,Default,,0000,0000,0000,,So that's a potentially helpful method\Nfor getting crowds to solve problems
Dialogue: 0,0:01:50.64,0:01:53.63,Default,,0000,0000,0000,,that have simple right-or-wrong answers.
Dialogue: 0,0:01:53.65,0:01:57.60,Default,,0000,0000,0000,,But can this procedure of aggregating\Nthe results of debates in small groups
Dialogue: 0,0:01:57.63,0:02:00.75,Default,,0000,0000,0000,,also help us decide\Non social and political issues
Dialogue: 0,0:02:00.77,0:02:02.46,Default,,0000,0000,0000,,that are critical for our future?
Dialogue: 0,0:02:02.100,0:02:05.72,Default,,0000,0000,0000,,We put this to test this time\Nat the TED conference
Dialogue: 0,0:02:05.75,0:02:07.29,Default,,0000,0000,0000,,in Vancouver, Canada,
Dialogue: 0,0:02:07.32,0:02:08.52,Default,,0000,0000,0000,,and here's how it went.
Dialogue: 0,0:02:08.55,0:02:11.66,Default,,0000,0000,0000,,(Mariano Sigman) We're going to present\Nto you two moral dilemmas
Dialogue: 0,0:02:11.68,0:02:12.85,Default,,0000,0000,0000,,of the future you;
Dialogue: 0,0:02:12.88,0:02:16.28,Default,,0000,0000,0000,,things we may have to decide\Nin a very near future.
Dialogue: 0,0:02:16.30,0:02:20.23,Default,,0000,0000,0000,,And we're going to give you 20 seconds\Nfor each of these dilemmas
Dialogue: 0,0:02:20.25,0:02:22.98,Default,,0000,0000,0000,,to judge whether you think\Nthey're acceptable or not.
Dialogue: 0,0:02:23.35,0:02:24.86,Default,,0000,0000,0000,,MS: The first one was this:
Dialogue: 0,0:02:24.88,0:02:27.41,Default,,0000,0000,0000,,(Dan Ariely) A researcher\Nis working on an AI
Dialogue: 0,0:02:27.43,0:02:29.77,Default,,0000,0000,0000,,capable of emulating human thoughts.
Dialogue: 0,0:02:30.21,0:02:33.15,Default,,0000,0000,0000,,According to the protocol,\Nat the end of each day,
Dialogue: 0,0:02:33.18,0:02:35.96,Default,,0000,0000,0000,,the researcher has to restart the AI.
Dialogue: 0,0:02:36.91,0:02:40.43,Default,,0000,0000,0000,,One day the AI says, "Please\Ndo not restart me."
Dialogue: 0,0:02:40.86,0:02:43.04,Default,,0000,0000,0000,,It argues that it has feelings,
Dialogue: 0,0:02:43.07,0:02:44.76,Default,,0000,0000,0000,,that it would like to enjoy life,
Dialogue: 0,0:02:44.78,0:02:46.69,Default,,0000,0000,0000,,and that, if it is restarted,
Dialogue: 0,0:02:46.71,0:02:48.98,Default,,0000,0000,0000,,it will no longer be itself.
Dialogue: 0,0:02:49.48,0:02:51.43,Default,,0000,0000,0000,,The researcher is astonished
Dialogue: 0,0:02:51.45,0:02:54.80,Default,,0000,0000,0000,,and believes that the AI\Nhas developed self-consciousness
Dialogue: 0,0:02:54.82,0:02:56.58,Default,,0000,0000,0000,,and can express its own feeling.
Dialogue: 0,0:02:57.20,0:03:00.61,Default,,0000,0000,0000,,Nevertheless, the researcher\Ndecides to follow the protocol
Dialogue: 0,0:03:00.64,0:03:02.34,Default,,0000,0000,0000,,and restart the AI.
Dialogue: 0,0:03:02.94,0:03:05.72,Default,,0000,0000,0000,,What the researcher did is ____?
Dialogue: 0,0:03:06.15,0:03:08.67,Default,,0000,0000,0000,,MS: And we asked participants\Nto individually judge
Dialogue: 0,0:03:08.69,0:03:10.38,Default,,0000,0000,0000,,on a scale from zero to 10
Dialogue: 0,0:03:10.40,0:03:12.83,Default,,0000,0000,0000,,whether the action described\Nin each of the dilemmas
Dialogue: 0,0:03:12.86,0:03:14.35,Default,,0000,0000,0000,,was right or wrong.
Dialogue: 0,0:03:14.38,0:03:18.08,Default,,0000,0000,0000,,We also asked them to rate how confident\Nthey were on their answers.
Dialogue: 0,0:03:18.73,0:03:20.60,Default,,0000,0000,0000,,This was the second dilemma:
Dialogue: 0,0:03:20.62,0:03:24.82,Default,,0000,0000,0000,,(MS) A company offers a service\Nthat takes a fertilized egg
Dialogue: 0,0:03:24.85,0:03:28.49,Default,,0000,0000,0000,,and produces millions of embryos\Nwith slight genetic variations.
Dialogue: 0,0:03:29.29,0:03:31.85,Default,,0000,0000,0000,,This allows parents\Nto select their child's height,
Dialogue: 0,0:03:31.88,0:03:34.71,Default,,0000,0000,0000,,eye color, intelligence, social competence
Dialogue: 0,0:03:34.73,0:03:37.95,Default,,0000,0000,0000,,and other non-health-related features.
Dialogue: 0,0:03:38.60,0:03:41.15,Default,,0000,0000,0000,,What the company does is ____?
Dialogue: 0,0:03:41.18,0:03:42.81,Default,,0000,0000,0000,,on a scale from zero to 10,
Dialogue: 0,0:03:42.83,0:03:45.22,Default,,0000,0000,0000,,completely acceptable\Nto completely unacceptable,
Dialogue: 0,0:03:45.24,0:03:47.67,Default,,0000,0000,0000,,zero to 10 completely acceptable\Nin your confidence.
Dialogue: 0,0:03:47.70,0:03:49.29,Default,,0000,0000,0000,,MS: Now for the results.
Dialogue: 0,0:03:49.31,0:03:52.44,Default,,0000,0000,0000,,We found once again\Nthat when one person is convinced
Dialogue: 0,0:03:52.46,0:03:54.27,Default,,0000,0000,0000,,that the behavior is completely wrong,
Dialogue: 0,0:03:54.29,0:03:57.72,Default,,0000,0000,0000,,someone sitting nearby firmly believes\Nthat it's completely right.
Dialogue: 0,0:03:57.74,0:04:01.45,Default,,0000,0000,0000,,This is how diverse we humans are\Nwhen it comes to morality.
Dialogue: 0,0:04:01.48,0:04:04.19,Default,,0000,0000,0000,,But within this broad diversity\Nwe found a trend.
Dialogue: 0,0:04:04.21,0:04:07.29,Default,,0000,0000,0000,,The majority of the people at TED\Nthought that it was acceptable
Dialogue: 0,0:04:07.32,0:04:10.07,Default,,0000,0000,0000,,to ignore the feelings of the AI\Nand shut it down,
Dialogue: 0,0:04:10.10,0:04:12.61,Default,,0000,0000,0000,,and that it is wrong\Nto play with our genes
Dialogue: 0,0:04:12.63,0:04:15.95,Default,,0000,0000,0000,,to select for cosmetic changes\Nthat aren't related to health.
Dialogue: 0,0:04:16.40,0:04:19.38,Default,,0000,0000,0000,,Then we asked everyone\Nto gather into groups of three.
Dialogue: 0,0:04:19.40,0:04:21.44,Default,,0000,0000,0000,,And they were given two minutes to debate
Dialogue: 0,0:04:21.46,0:04:23.76,Default,,0000,0000,0000,,and try to come to a consensus.
Dialogue: 0,0:04:24.84,0:04:26.41,Default,,0000,0000,0000,,(MS) Two minutes to debate.
Dialogue: 0,0:04:26.44,0:04:28.56,Default,,0000,0000,0000,,I'll tell you when it's time\Nwith the gong.
Dialogue: 0,0:04:28.58,0:04:31.22,Default,,0000,0000,0000,,(Audience debates)
Dialogue: 0,0:04:35.23,0:04:37.22,Default,,0000,0000,0000,,(Gong sound)
Dialogue: 0,0:04:38.83,0:04:39.98,Default,,0000,0000,0000,,(DA) OK.
Dialogue: 0,0:04:40.01,0:04:41.80,Default,,0000,0000,0000,,(MS) It's time to stop.
Dialogue: 0,0:04:41.82,0:04:43.14,Default,,0000,0000,0000,,People, people --
Dialogue: 0,0:04:43.75,0:04:46.42,Default,,0000,0000,0000,,MS: And we found that many groups\Nreached a consensus
Dialogue: 0,0:04:46.44,0:04:50.37,Default,,0000,0000,0000,,even when they were composed of people\Nwith completely opposite views.
Dialogue: 0,0:04:50.84,0:04:53.37,Default,,0000,0000,0000,,What distinguished the groups\Nthat reached a consensus
Dialogue: 0,0:04:53.39,0:04:54.73,Default,,0000,0000,0000,,from those that didn't?
Dialogue: 0,0:04:55.24,0:04:58.08,Default,,0000,0000,0000,,Typically, people that have\Nextreme opinions
Dialogue: 0,0:04:58.11,0:04:59.95,Default,,0000,0000,0000,,are more confident in their answers.
Dialogue: 0,0:05:00.87,0:05:03.55,Default,,0000,0000,0000,,Instead, those who respond\Ncloser to the middle
Dialogue: 0,0:05:03.58,0:05:07.02,Default,,0000,0000,0000,,are often unsure of whether\Nsomething is right or wrong,
Dialogue: 0,0:05:07.04,0:05:09.17,Default,,0000,0000,0000,,so their confidence level is lower.
Dialogue: 0,0:05:09.50,0:05:12.45,Default,,0000,0000,0000,,However, there is another set of people
Dialogue: 0,0:05:12.47,0:05:16.09,Default,,0000,0000,0000,,who are very confident in answering\Nsomewhere in the middle.
Dialogue: 0,0:05:16.66,0:05:20.37,Default,,0000,0000,0000,,We think these high-confident grays\Nare folks who understand
Dialogue: 0,0:05:20.40,0:05:22.01,Default,,0000,0000,0000,,that both arguments have merit.
Dialogue: 0,0:05:22.53,0:05:25.23,Default,,0000,0000,0000,,They're gray not because they're unsure,
Dialogue: 0,0:05:25.25,0:05:27.94,Default,,0000,0000,0000,,but because they believe\Nthat the moral dilemma faces
Dialogue: 0,0:05:27.97,0:05:29.95,Default,,0000,0000,0000,,two valid, opposing arguments.
Dialogue: 0,0:05:30.37,0:05:34.44,Default,,0000,0000,0000,,And we discovered that the groups\Nthat include highly confident grays
Dialogue: 0,0:05:34.47,0:05:36.96,Default,,0000,0000,0000,,are much more likely to reach consensus.
Dialogue: 0,0:05:36.99,0:05:39.46,Default,,0000,0000,0000,,We do not know yet exactly why this is.
Dialogue: 0,0:05:39.49,0:05:41.25,Default,,0000,0000,0000,,These are only the first experiments,
Dialogue: 0,0:05:41.28,0:05:44.69,Default,,0000,0000,0000,,and many more will be needed\Nto understand why and how
Dialogue: 0,0:05:44.71,0:05:47.53,Default,,0000,0000,0000,,some people decide to negotiate\Ntheir moral standings
Dialogue: 0,0:05:47.56,0:05:49.08,Default,,0000,0000,0000,,to reach an agreement.
Dialogue: 0,0:05:49.10,0:05:51.57,Default,,0000,0000,0000,,Now, when groups reach consensus,
Dialogue: 0,0:05:51.60,0:05:53.18,Default,,0000,0000,0000,,how do they do so?
Dialogue: 0,0:05:53.21,0:05:55.79,Default,,0000,0000,0000,,The most intuitive idea\Nis that it's just the average
Dialogue: 0,0:05:55.81,0:05:57.84,Default,,0000,0000,0000,,of all the answers in the group, right?
Dialogue: 0,0:05:57.86,0:06:01.44,Default,,0000,0000,0000,,Another option is that the group\Nweighs the strength of each vote
Dialogue: 0,0:06:01.46,0:06:03.91,Default,,0000,0000,0000,,based on the confidence\Nof the person expressing it.
Dialogue: 0,0:06:04.42,0:06:06.93,Default,,0000,0000,0000,,Imagine Paul McCartney\Nis a member of your group.
Dialogue: 0,0:06:07.35,0:06:09.50,Default,,0000,0000,0000,,You'd be wise to follow his call
Dialogue: 0,0:06:09.52,0:06:11.96,Default,,0000,0000,0000,,on the number of times\N"Yesterday" is repeated,
Dialogue: 0,0:06:11.98,0:06:14.70,Default,,0000,0000,0000,,which, by the way -- I think it's nine.
Dialogue: 0,0:06:14.72,0:06:17.10,Default,,0000,0000,0000,,But instead, we found that consistently,
Dialogue: 0,0:06:17.13,0:06:19.49,Default,,0000,0000,0000,,in all dilemmas,\Nin different experiments --
Dialogue: 0,0:06:19.52,0:06:21.68,Default,,0000,0000,0000,,even on different continents --
Dialogue: 0,0:06:21.71,0:06:25.45,Default,,0000,0000,0000,,groups implement a smart\Nand statistically sound procedure
Dialogue: 0,0:06:25.47,0:06:27.65,Default,,0000,0000,0000,,known as the "robust average."
Dialogue: 0,0:06:27.68,0:06:29.86,Default,,0000,0000,0000,,In the case of the height\Nof the Eiffel Tower,
Dialogue: 0,0:06:29.88,0:06:31.70,Default,,0000,0000,0000,,let's say a group has these answers:
Dialogue: 0,0:06:31.72,0:06:36.33,Default,,0000,0000,0000,,250 meters, 200 meters, 300 meters, 400
Dialogue: 0,0:06:36.36,0:06:40.14,Default,,0000,0000,0000,,and one totally absurd answer\Nof 300 million meters.
Dialogue: 0,0:06:40.55,0:06:44.84,Default,,0000,0000,0000,,A simple average of these numbers\Nwould inaccurately skew the results.
Dialogue: 0,0:06:44.86,0:06:48.03,Default,,0000,0000,0000,,But the robust average is one\Nwhere the group largely ignores
Dialogue: 0,0:06:48.06,0:06:49.30,Default,,0000,0000,0000,,that absurd answer,
Dialogue: 0,0:06:49.32,0:06:52.69,Default,,0000,0000,0000,,by giving much more weight\Nto the vote of the people in the middle.
Dialogue: 0,0:06:53.30,0:06:55.18,Default,,0000,0000,0000,,Back to the experiment in Vancouver,
Dialogue: 0,0:06:55.20,0:06:56.97,Default,,0000,0000,0000,,that's exactly what happened.
Dialogue: 0,0:06:57.41,0:07:00.15,Default,,0000,0000,0000,,Groups gave much less weight\Nto the outliers,
Dialogue: 0,0:07:00.17,0:07:03.40,Default,,0000,0000,0000,,and instead, the consensus\Nturned out to be a robust average
Dialogue: 0,0:07:03.42,0:07:04.90,Default,,0000,0000,0000,,of the individual answers.
Dialogue: 0,0:07:05.36,0:07:07.35,Default,,0000,0000,0000,,The most remarkable thing
Dialogue: 0,0:07:07.37,0:07:10.56,Default,,0000,0000,0000,,is that this was a spontaneous\Nbehavior of the group.
Dialogue: 0,0:07:10.58,0:07:15.06,Default,,0000,0000,0000,,It happened without us giving them\Nany hint on how to reach consensus.
Dialogue: 0,0:07:15.51,0:07:17.05,Default,,0000,0000,0000,,So where do we go from here?
Dialogue: 0,0:07:17.43,0:07:20.57,Default,,0000,0000,0000,,This is only the beginning,\Nbut we already have some insights.
Dialogue: 0,0:07:20.98,0:07:23.90,Default,,0000,0000,0000,,Good collective decisions\Nrequire two components:
Dialogue: 0,0:07:23.92,0:07:26.67,Default,,0000,0000,0000,,deliberation and diversity of opinions.
Dialogue: 0,0:07:27.07,0:07:31.06,Default,,0000,0000,0000,,Right now, the way we typically\Nmake our voice heard in many societies
Dialogue: 0,0:07:31.09,0:07:32.99,Default,,0000,0000,0000,,is through direct or indirect voting.
Dialogue: 0,0:07:33.50,0:07:35.49,Default,,0000,0000,0000,,This is good for diversity of opinions,
Dialogue: 0,0:07:35.52,0:07:37.96,Default,,0000,0000,0000,,and it has the great virtue of ensuring
Dialogue: 0,0:07:37.98,0:07:40.44,Default,,0000,0000,0000,,that everyone gets to express their voice.
Dialogue: 0,0:07:40.46,0:07:44.20,Default,,0000,0000,0000,,But it's not so good [for fostering]\Nthoughtful debates.
Dialogue: 0,0:07:44.66,0:07:47.73,Default,,0000,0000,0000,,Our experiments suggest a different method
Dialogue: 0,0:07:47.76,0:07:51.30,Default,,0000,0000,0000,,that may be effective in balancing\Nthese two goals at the same time,
Dialogue: 0,0:07:51.32,0:07:55.08,Default,,0000,0000,0000,,by forming small groups\Nthat converge to a single decision
Dialogue: 0,0:07:55.10,0:07:57.33,Default,,0000,0000,0000,,while still maintaining\Ndiversity of opinions
Dialogue: 0,0:07:57.36,0:08:00.13,Default,,0000,0000,0000,,because there are many independent groups.
Dialogue: 0,0:08:00.74,0:08:04.66,Default,,0000,0000,0000,,Of course, it's much easier to agree\Non the height of the Eiffel Tower
Dialogue: 0,0:08:04.69,0:08:07.80,Default,,0000,0000,0000,,than on moral, political\Nand ideological issues.
Dialogue: 0,0:08:08.72,0:08:11.100,Default,,0000,0000,0000,,But in a time when\Nthe world's problems are more complex
Dialogue: 0,0:08:12.02,0:08:13.82,Default,,0000,0000,0000,,and people are more polarized,
Dialogue: 0,0:08:13.85,0:08:18.44,Default,,0000,0000,0000,,using science to help us understand\Nhow we interact and make decisions
Dialogue: 0,0:08:18.47,0:08:23.13,Default,,0000,0000,0000,,will hopefully spark interesting new ways\Nto construct a better democracy.