Speech Transcript - Collective Intelligence and Team Learning (July 2014)

of the different kind of all preamble anyway

a little bit preamble we also know from that same study that

you know when you do you have high functioning teams

those schemes can generate the last and second you propagated other teams and you know

teams

we have a fair amount of data now in the

and the organisational behaviour literature more generally that teams can do a fair amount of

like areas learning from other teams so

really in here enhancing the quality of

teams at the team level can generate benefits for the plateau organisation

and finally and in some work that and doing that it will be describing today

but another line of work on the board and which includes michael o'leary a georgetown

i don't know with

are georgetown colleague here does michael but

we have been looking at multiple t membership and how you know really fat the

organisation scanned

strategically manage the assignment of people across scheme so that those ideas can cross pollen

eight across projects and they'll

that's all today that a lot of wanting and we talking about today at the

team level but i think that there are ways that we can read from that

the thinking about broader organisational by fax the of improving team collaboration

so without let me tell you a bit about the work within doing on collective

intelligence

in some gonna start with examples from the animal kingdom if in any of you

been familiar with some of the word my colleagues or i present you have to

do start here because the

because there isn't really fascinating example that i think that there are

a few observations we can generalize well

until recently if you had global collective intelligence than in even now

a lot of the are really hits you'll get in your search are examples from

the animal kingdom and in fact

we now have a collective intelligent conference

that's held every year or so at mit

and menu to people come present our people just study

these ideas any animals

because they're very they're really fascinating examples price to learn from in general i us

to it's for example are very you know simple creatures individually they don't have a

whole lotta

memory they don't have a whole lot of problem solving capacity but collectively because of

the effectiveness with which they coordinate

they can accomplish i huge range of things they can

build structures their fairly complex they can build bridges to traverse to rain that they

could and individually they can't hear e things

that are many multiple of their body way all because they have some very effective

mechanisms for coordinating individual after

into

we know one way in which these

examples don't generalized has to do with the range of things they try to accomplish

you know neither you know animal examples

there's a very limited

range of wave in which they're trying to coordinate

but they are hardwired to accomplish that coordination effectively sell and the an example

their fair amount trail today we

other in its follow the strength of the trail you know indicates to them how

many in the got along a particular path or you know whether or not it's

a good knoll the signal phase over time

and so

enables them to or me quite effectively in so we started to wonder if there

are groups

that i engage in a very high level high quality collaboration that consistent over time

and across domains

that essentially enables them to perform well at least in the range of things that

they're doing as a group

you know in that context in so

we know that there are some groups that perform at a very high level in

particular domain so this is a picture of the wealthiest chamber orchestra

there are world renowned chamber orchestra

that is also well known for being either a lesser conductor let they don't have

that

you know and appointee conductor they rotate model of shared leadership on the different pieces

that they play

so we know that or if yes excels and the music domain

but we wondered in our research is whether or not there is a measurable collective

intelligence teams that would differentiate your seventeens from others

and

and would give us insight into the prosody that they used to collaborate correlate effectively

over a wide range of situations and over time and can we use that information

actually project on which schemes are likely to perform well in the future so that

where we started this work

in starting this work we were really drawing on

ideas about general intelligence that originated with charles german

individual intelligent or i q as many

instruments refer to it you know originated with your men's work in the early twentieth

century he observe that school children who were good in one academic topic tended to

be also very good in and many other academic subjects

anything about

to where the empirically

and

in the va the really precursors a factor analysis which those who do research in

psychology and related domains are probably familiar with

but what he did of the isolated the underlying factor individual a factor of intelligent

orgy factor

and what we know that the chief that are that underlies performance in many domain

and this is the thing that individual like you task actually or task in the

level of g for an individual in so what we know from now decades and

actually over sentry of research after experiments initial work

is that individual intelligence

measured by task that can be administered in a short amount of time can predict

many outcomes over a broad range of domains over long periods of time so

if you give school children and intelligence tasks that not only projects how the performance

but also their career six that the probability with which to become an unwed teenage

period

i mean even their life expectancy so many domains over long periods of time

so we wondered if the same thing would be true for teams is there a

general collective intelligence factor or three factor to parallel a g fact there which day

just the degree to which

a just the quality of collaboration in coordination in a key and predict how well

they're likely to perform in the future

can you measure that their way to measure it can we predict which teams are

likely to perform well especially in advance of some fairly high speech projects their situations

may might put "'em" and i actually talked about this way to dump looks at

nasa who are interested in thinking about this with their groups of ask or not

because of the disastrous consequences

that they can experience if people

don't work well up on the space station and this is tell us something beyond

knowing individual intelligence so

in two we initiated this work a few years ago when we had talked about

intelligent teams it was as a function of the individual intelligence of numbers

and there is a lot well the research looking at you know what is the

relationship between average individual intelligence

indy performance and the

the answer is kind of complicated it depends on a lot of factors we wondered

of collective intelligence if we could measure a

we tell at something beyond individual intelligence with the is that a unique

capability that rim resided the group level that is the gauge of the effectiveness of

how the individuals work together

so we thought about doing that the research on methanol just briefly tell you about

our findings

we brought groups into the lab for several hours they worked on a whole range

of tasks we found strong evidence of an underlying collective intelligence factor

and that factor was a strong predictor of performance at a later point and end

time or more complex task multifaceted task in this case with the video games simulation

we have them deal collective intelligence with a much stronger predictor your performance

then the average i q or numbers or

many people ask what about the smartest person also but also to predict very much

about how the team performed especially compared to collective intelligence we replicated in this with

a broader with a different set of tasks a broader range of group sizes

i different outcome task and again found strong evidence of that

of a key factor

indian performance that predicted how these schemes

performed on this more complex task in much more cell

then the intelligence of individual members

and so just to kind of summarize the relative value of individual intelligence and collective

intelligence the dark bars are collective intelligence this is the pretty value of average individual

member intelligent and this is the value of

the maximum or the smartest person

in the group and so what we see is that the collective intelligence factor actually

gives us pretty good ability to predict which groups are likely to perform well in

the feature and in which all right

maybe more germane perhaps the thumb of the things will be talking about today we

also looked at that in the context of learning so we've now got nine and

several studies since our original paper command science a few years ago and looked at

it in a number of contexts predicting a number of different things over you no

longer ranges of time

and in found you know fairly consistent effects here where we're also interested in morning

in this particular task what we're getting at is what

some call implicit learning or implicitly coordination so this is the task where the group

play multiple rounds of again

where group members were paid on the basis of what they decided but also what

others in the group decided and these decisions were made with out communication so they

are on a computer

they're looking at a matrix

and they can i should of right a an example output of an example of

the matrix but basically

all the group members make a choice from the matrix and their p the on

the basis of the combination of choices that different group members make again without

communication and so

you know the start of in the first round after each round they find out

what their payoff wide and then they can make a choice for the next row

and basically what they're doing is they're trying to make

aghast about what others in the group are likely to choose based on you know

that outcomes that they're seeing with each round

itself

before they did if game we

gave them are caff battery to measure are their collective intelligence

and what's interesting about this is that in the first round

so the red line here are the groups that high collective intelligence the orange line

or yellow line or groups that were blowing collective intelligence

in what we see that the very first round groups that are high and low

collective intelligence don't really you know have exhibit very large differences in their performance in

fact if anything the high highly collectible intelligent groups are a little lower well as

you follow this red line you see that over the series of rounds

the highly collectively intelligent groups get better battery there's kind of this fairly steady climb

the and that significantly higher than where they began

overall

the members are much more money

then the group and the one

well collective intelligence groups where

it kinda jump around but they and that approximately where they began

and the are and significantly less money overall

in so in the economics literature these games as i mentioned are interpreted as gauging

implicit coordination and

and what we see is that there's something about the way that the groups that

are highly collective intelligence or functioning that enables them to sort of intensity what other

group members might be doing adjust their behaviour accordingly and perform better overall is the

you met

which we finally interesting

in another study this one was done in a classroom setting

again we gave teams beginning these routines that were

randomly assigned to work together as a group over the course of the semester we

gave them are collective intelligence battery at the beginning of the semester and then in

this particular course the teams had to take

i theories of exams together

and the first took the ml than individual and then without getting feedback on how

they done they did the same set of exam problems as a group

in this is part of the t f t vs learning if anybody's familiar with

that approach

it's two

again we measured collective intelligence the beginning of the term what we sat was that

the groups that were high collective intelligent shared steady improvement over the course of the

exams and the data together these are the teams choirs

a groups that were lower collective intelligence again not a not exactly upward trend but

without would be groups that will hire

but most interesting for us with even

how the group data relative to their best member so in this case of the

imagine a group the individual members to begin by themselves

and we were interested in how the group the relative to the student that in

the past an exam before the group get it together and what we see is

that and the groups that a high collective intelligence the group so we score significantly

better than their best number

whereas in the groups there alone collective intelligence their best members are just as good

as the members in the highly collectible intelligent teams but the group is not always

or better than that of course

in fact you know in the second exam a or

slightly worse

in so

there's something about the we that these groups that are high in collective intelligence or

coordinating that enables down to fill in the gaps of members knowledge so that they

together can actually give better an exam than any individual dead

on there are

for some reason that groups there alone collective intelligence this doesn't

so very told you that we try to predict collective intelligence on the basis of

individual intelligent and the correlation was very low so that kind of brings about the

question what does predict collective intelligence in so they were number of things that were

surprising that as we've been exploring

so the first is that it's not really directly predictable by the quality of relationships

that group members they're experiencing in the group and then we give invalidated scales of

things like good satisfaction and group cohesion and

of motivation et cetera and generally find no correlation between what group member say about

their experience and the matter collective intelligence of the group

this is not necessarily what they really the case that a group that deals more

group b is actually more collectively intelligent

in when i presented these results to a few different audiences in the military especially

the army in particular

they really wonder about the cohesion thing because there is a really strong belief there

that cohesion is the basis all other you know good performance variables and so this

research kind of call that are quite question

we've done a range of analyses on different personality managers

you know including the that

com and big five and related instruments of the main personality variables

it down no consistent relationship

with any sort of

purpose personality configurations in collective intelligence

one thing that we have found

you know fairly consistently and the relationship of the proportion of female in the group

in so

we recently kind of abrogated our data across several studies and so this represent hundreds

of groups

in what we find that when you're and so what we did we classify each

study on the basis of whether or not you know the corpus all male they

just had a single female

if it was majority male they had more than one female that it would be

in this point rather than in this at this point

if you know with half and half and then on over to a female and

what we feel that it is zero is the mean collective intelligence to me collective

intelligence is kind of a standard i score and so

zero is need

and here we use the is that it's when you get a majority female

or especially when you just have one now that groups more consistently reach

the highest level the collective intelligence when you get the all female

you see that the mean here draw the there are some benefit to diversity

but one that you know fevers female verses you know if you could be

so why we've continue to explore a little bit about why that might be and

you know all cover a few more results here let's start to give us a

little bit more insight

and so one of them have to do with inability that summer for the social

perceptive the sometimes i'm social intelligence

to in our study people completed a measure called the reading the mind in the

i have in this was originally developed by researchers of octave them and autism spectrum

disorders and so it involves looking at thirty six pictures of the eye region of

the and using

which of these options

you think represents what this person is thinking or feeling

in so

it turns out that if is very difficult for people who of art is ever

are not doesn't spectrum disorder

and it turns out that it also difficult for people and then normal and of

the continuum as well

and women on average or slightly better on this measure them and that was true

added into an all of our sample i mean it's true in the population overall

and this turns out to be highly predictable of collective intelligence the mean on this

scale

and so disprove this explained a lot of the f that not completely but a

lot of the in fact of the proportion of women in the group of collective

intelligence that you actually raise the average level of social perceptive

another compositional thing that we've looked at recently we start to look at various forms

of diversity

and so one form of diversity that

one of my graduate students is particularly interested in cognitive style diversity

and so what she finds of that this it's this moderate level of cognitive diversity

you see that there is curve a linear relationship but at the moderate level code

of cognitive diversity that seems to most enhance collective intelligence

and in similar for where recent study she's been looking a lot at you know

what exactly if that makes a group only moderately cognitively diverse and what you find

is that the existence of individuals you actually can

decline multiple different cognitive style

or one of the most important things that are present more often than is moderately

cognitively diverse groups so somebody you can take the perspective of

that is that what we're call it what the styles are called but just

to simplify for the moment you take mormon engineering perspective

but then also take more of an artistic perspective on a problem can help people

who are adapting only one cognitive style understand each other batter

and if is consistent with some other studies that are coming out on more recently

in the literature some other colleagues of mine have died

i'll looking at multicultural of them

p or people who are bilingual

and the role that a plane facilitating the group to their part have

in sort of helping bridge the gap between you know to people who anyways wouldn't

understand each other and they were finding that having some of these focus groups seems

to you know also enhance collective intelligence

so i've talked you about a set of things that you know we kind of

think of the screen compositional variables these are things that if you could screen people

in compose the ideal group

you know these might be things that you with pay attention to

but sometime you don't have that lingerie and so they're than other sort of things

that we've looked at

that have to do more with how the group actually behaves are interacts one there

together

it's the one set of findings really relates to group communication

so in our in our initial study we measured smoking behavior and patterns of speaking

and interrupting so our group members all war the gizmos called social badges they were

invaded by one of our collaborators at mit

see any pentland is written a lot about these in various contexts

but in our case we had them wearing i'm so that we could pick up

patterns of spoken language

and what we found is that

sort of more or distribution of speaking turns or

i was most strongly associated with high collective intelligence or in other words of somebody

was dominating the conversation and doing all of the talking kind of like i am

right now if this is how our whole interaction one that we will be very

close intelligent probably

and that's what we find a in our study

in we've also found it even with that's one studies that we dined with online

and so

we know that some studies where groups are interacting just by a chat

and we find that

you know total amount of communication and is predictive of collective intelligence but also equality

of communication

in this even generalize is over into now when we measure collective intelligence we use

a platform where we can keep track of

applying the answers to questions are battery and whose you know doing the work of

the tasks that we're having the group or and still we also can measure the

equality contribution to that in we all we find i think that or equal contribution

to the work of the group

is also associated with higher clock intelligence

in so then another piece that i'm gonna speech to on this is sort of

is that we talk about level communication but and it also how and what are

at communicating about in here we've been looking at levels of collaboration and integration

in still in a in a study that we published if you years ago now

we were giving groups basically care is

case scenario to solve in we compose these groups so that they either are included

some key pieces of expertise among them so experts in the topics that they needed

to analyzed in order to solve the case or they get in

but we also had i their spending time at the beginning of their work really

thinking through the aspects of the problem and how they should integrate the analyses that

they were gonna be doing individually or we get and we just let them kind

of start the way that they normally word

and what we found that when the group didn't have the expertise they needed

i didn't matter if we gave them some way of integrating this expertise you know

in this kind a higher quality weighted in how the information they needed to make

a release all the case

but more interesting was what happened when we did have the experts in the group

we have experts in the group but we didn't have them

spend time thinking about how to integrate their work the group actually perform significantly worse

than that they didn't have the expert

and what we found is that these group members work even more in isolation than

the groups that were lacking the expertise because they sort of this unit each person

kinda knew what they needed to do you do their partners together right in the

group would be fine so they didn't actually collaborate integrate

as much as the as the groups and the other condition however when the group

have the expertise in the and we gave them again a fairly simple

sort of starting exercise to think about how to integrate their work you know they

perform the basketball

in so it you don't really suggests that has that it's not about how you

compose the group of making sure you have the right people in it but what

they actually still when they're together

and when i tell people about some of these findings

i often dry analogies with madison

and i'm a big fan of a lot of what had to go one the

right in his but about the medical profession but one point he takes than the

others make it you know

in that it then they're doing a lot of research to try to come up

with bigger and better way to do things but in fact there are a lot

of really simple things that they are you know have huge impact in this trouble

is really just to get groups to do it to get people to do it

and so one really common intervention for example hospital if hand washing

right

we know that hand washing is one of the thing almost important things that

medical professionals can do this but to control the spread of the action however you

know it often doesn't happen in the then they have to do this big interventions

get people to last

and so another simple thing that

you know was tried a few years ago and had a huge tax is that

surgery chocolate and essentially it sounds very much like the type of thing we did

with our teams

be in the lab study

in this case they have a checklist again through the make sure the introduce themselves

they make sure they know what keys they're working on and you know where they're

doing their operation and they make sure they have all of the supplies that they

need it's that are tracked that are down the list and they found fairly large

for this fairly simple intervention because you know i'm sure many but support for stories

of people having the wrong one cutoff or whatever you know just because them fairly

key details or not from the beginning and sell everything is that you know what

our study suggest

well two things one is that i know from doing mathematically studies now then when

you bring a team into the room if you don't make them stop

and sorta

introduce themselves and use "'em" key things they don't

and it makes a huge difference in terms of how they operate as a team

and i think that's true and a lot of organisational settings as well and so

they're probably from fairly simple interventions that can be in a

the help teens achieve higher levels of integration

it's a while

we're still doing research to find even bigger fancier ways for them to achieve integration

and i think the implementation of the simple things well while but that

and then the final thing and this is like hot off the path in fact

we just present these results that a conference last week and my graduates didn't you

get the work on that's actually one an award for the poster

on this so you're one of the first groups that i'm gonna tell about it

but it really to trying to reduce as much as possible the intra groups that

of competition

and the effect that has for collective intelligence so in this particular study

we composed groups to include between zero in four when n

and then we had three different conditions we had sort of the condition where we

didn't do anything to manipulate status of group members

we can condition in which

we had them a lack the reader

and then after condition in which we had the molecular either but we told that

at the end of the

or later in the fashion name i have a chance to electoral a different year

if they want to deal

and what we found with that

that condition where the elected the we heard and they could change that meter is

what elicited the most

that its computation and so we later ask group members about their experiences that of

competition

in the group

and so what is in this graph is that if static computation on their measured

bubble of collective intelligence

as a function of having women are in the group

and what you see if that so that the solid line if the highest levels

of static competition

and basically at that of competition goes up

and the number of women in the group goes up

collective intelligence goes away down

and we know that's added competition is the highest in those conditions in which the

that the leader could change

whereas when status competition is low and especially as number of women in the group

go up paper for a whole lot at

and so we're still

looking at the nature of the information sharing the happen in the group you know

how that of competition was an active

a basically

you know kind of a really sense that i think people to the difference that

of the members of the group giving a sense that no one person could kind

of overtaken other in viewership of the group seems to have a really does more

facts for collective intelligence in groups and especially

as more women are included and

i be talking a lot recently i've been called by groups that are interested integrating

one and into their profession so you know engineering society vehicle society they're to that

seem to deal

especially focusing on that

and you know part of it is maybe when there is a lot of attention

test data in a particular profession women don't even want to be there to begin

what

and so this is something that you know with tax that a lot about is

you know to the degree that it's a very static on chat sort of setting

you know if that one of the things that is keeping one and from wanting

to stay

in with no certainly even if they use a on it doesn't help the quality

of the work of the groups that there

so another important feature not necessarily one that easy to resolve and reaching back to

the medical example with this is there in spades in terms of the status differences

that are very apparent in a lot of medical setting something that they reluctantly

trying to think about grapple with as well

so just by way of kind of summarising in wrapping up here

we now done you know dozens of studies looking at the existence of c factor

in crude and you know keep finding evidence of it in a wide variety of

settings so it's just underlying capability that captures the quality of collaboration in coordination in

a group

i that seems to be relatively stable absent some big intervention and

into et

and predicts future performance as well as group learning and the inability programmers to make

the most of what are the team to make the most of what individual members

now

and what we see what we look across the kind the factors that theme to

facilitate collective intelligence they all relate to

things that facilitate the transfer an integration of information

so people who are better at social perception or better able to pick up and

nonverbal cues in factor that into the way that they behave and how they

conduct group conversation

groups that are either low or moderate incarnated versa your have individuals in the group

who know how to read the different cues are the different perspectives of other members

giving more collectively intelligent

when we make sure everybody can be part of the conversation and relatively low

proportions we get more information from everybody more collect intelligence

interventions that enhance collaboration in integration are important and the status competition because as we

know some of the work of in the edmonton in others our business or

really looks that

you know how comfortable do all the different group members feel about speaking not in

a situation and if

that of competition high people are really paying attention is added then you probably a

certain people

we do not feel comfortable speaking up and that hurts the group

and that's very consistent what we found

so with that i'm going to close i

my domination of the conversation

and open it up i guess the questions a and other conversational

so now

hopefully o

i mean you can

any

i and then i

that and are higher than average and social perception so i suspect or okay

sure

that the great question

so we have you know so it's no longer the five or six hours in

the lab we've actually but for that you know had a big project of

creating it on my in a web based tool

where remembers login and so they can either be colocated are they can be at

different locations and a log and then

it gives them kind of a work study and in the you know facility that

they can use to communicate in the system kind of presents the problems that

and they can see each other's work as they work together kinda like google docks

if you've ever if not google docks exactly but it's that idea

and so and so that's what we used in so

the collective intelligence measures basically just had a the team perform across the different kinds

of problems we gave

and then go into a lot the problems that we gave

the groups but there are problems that

can all be done by us a little individual by themselves are really require different

kinds of collaboration across the group members so really gauge on the degree which they're

good at collaborating coordinating

that's interesting so that something we may be integrating into it at the moment the

way the system works of the day just have a limited amount of time to

work on each problem

and then we just take you know how well they perform a given that time

i'm in actually is well a lot of the standard i q task are structured

that way all the more of them are starting to also move toward the how

long to take approach to see if they can get a better a better estimate

that way

well

well as one okay well known for its right i well i l one

i star

a how i

so and making drying from analogy and i q task because of something that i

think a lot of people are familiar with

we really give them problems that don't require any particular domain expertise but are more

you know abstractions so that really what we're getting at is the group process verses

any particular content knowledge

actually i so same questions are you get a little bit different use and answered

it may be elaborate a little bit more and whether it only on a syllable

this what i think that i would have a actually the

i know it means for example we first apply prosody is that essentially and different

subject matter expertise is that what implies that

or is that something

so the kind of style diversity that we look at is beef and work still

you probably are familiar with the classic and educational distinction between people who are available

lasers of people who are visualise errors it more recent work and neuroscience the cognitive

psychology has for their distinguish different types of visualiser is there are some people who

are

object visualiser is their very good at they come up with very vivid mental images

to represent problems and kind of think about images holistic we in a tent were

quarry and sort of the visual arts a design

and their distinct emphatic terms kind of silly negative correlation

between the capabilities and of my calendars pop in your screen but

that there is a and negative correlation between the capabilities associated with that people who

are spatial visualiser is a really good a rotating objects and three dimensional space

and tend to be more represented in sort of engineering and that

professions and so

we use a manager that was developed by the cognitive psychologists you

identify this dimension and

and that classifies people a strong verbalize are strong but spatial visualiser is a strong

object visualise errors in there is a developing literature really

demonstrating how

people just engaging fundamentally different approaches to problem solving as a function of what their

dominant cognitive style as

and

you know so those are the dimension that we're using and so what we find

that diversity and groups create some problems although create some opportunities because those are associated

with different skills and so if you have people in the group we have the

skills obviously of the advantage of the scale but simply you know just talking together

about

how we wanna work together you know can be a little bit of a challenge

if it's a real eager brisk

i yes i scenes

brain clear

if you do i know if you higher

how he's and one

of course there are a well i schemes error bars

wow you please

and i hated this well this is with this study was the same

so no

well and so

so i be i should i have i'm familiar with seminar she's work on that

but i should look at what he

explored as in the explanation for why because

there's also this you know small but developing literature about the individuals were multi cultural

are multinational in the role that they can play

a disability not so i don't know if he has are people who are expat

do you can do the thinking and some care l

talk to everything and anything like that

well i e o

the question is a it's a well with the meeting area and

well

so you can

is it has been integration and

with initial work well i don't know and you met the some exercises duration

you look so well used or so

sure so i would say that it so to take the first question for so

cohesion is usually try to as

you know a concept that really to how people feel about the other people in

their group you know and so i know the latter the standard measures of cohesion

have to do with would you like to socialise with these people would you like

you know would you d

that is

these people left your group you have to do more with the interpersonal relationships with

integration is more of a cognitive

idea it's really more

do i see the connections between what i know in what i'm doing and what

you know in what you're doing

so that the distinction and under of that

make that are if i could be more if that doesn't

no i

and so

it so

in the exercises that we've use both in our lab stays but also i

what i suppose dark actually dead quite a bit work in the intelligence community and

that we were working with time analyses that were collaborating across agencies and so we

use similar a similar approach

but off then

in my case that a lot of the simple

has to do with having group members identify what their knowledge bases are

and then having the group together about what is the problem they're going to solve

the one of the parts of that problem and who would be passed to address

the different parts of the problem

and identifying what the overlaps are there so it's like okay you know brian and

i should be together about

international diversity in teams because you know we both know something about that i don't

know

the way you know i don't know about this but you know somebody else that

it just speaking systematically them at

a identify what numbers knowing be mapped onto the work that they're doing

in kind of a more explicit way

because otherwise what happens and i we thought it i observed a lot of exercising

the intelligence community before we started doing our work with the teens there

and you get to the end of the problem and they would discover

that there's this quiet woman in the corner who actually have the physics degree that

would have been really helpful for some of the stuff that we were looking at

and nobody in you know you know until the end and what was to eight

and that happens a lot and so i think just that simple step is you

know an initial

and initial interaction

i mean

as well

all right

and

as you were just going to do

so well

one

so i think you know it is really the path of like you know finding

out where f you have in how to put them together and what interesting so

you know that i remember that and that actually we did that would be intelligence

teens as well where you know if we were starting in echo

we have them spend five or ten minutes kare and interview each other

and then introduce your apartment or the group

and so that somebody brag about somebody held you know and how they're gonna be

really helpful and also you know create at least one you know more solid relationship

to start with nobody knows each other

and then let everybody else like you know you know what

what resources they have and the members

so i mean i again to downsample is if it is usually get really undermines

the ability of the team to really make use of what they have

well

you

one unsuccessful ones

one for people

we replace

right

used to

and

well

so we are starting to look at

at turn over

a bit in a different project that i didn't mention here

the

it depends on what the what the criteria of performance are

you know we find in some contexts where

you know what

generating you know deep insight or in our high levels of innovation

there is less harm from the

turnover but for the most power and you know i haven't come across anything at

that really suggests a great way to do deal with that i mean most of

the

the academic literature as well as you know the practical their of like you're mentioning

really just suggest that

you know turnover is

i had has a really deliver here if the fact

and when you

when you met your collective intelligence and eighteen am i think it can give you

actually an even better gauge of how much value you just way when you take

people out of the team or move them around

but unfortunately i mean i would love to come up a an organisation

who is dealing with that really well

because they have it

well is there i guess that is really

but the thing a more

at first

and

then

including

and you know

you know

all right

well

it's using the one

well i think i

well

i think it here

and

or not

i social

well as

we can work

there is thirty than another graduate student

is just completing and she with the my off as just the other day a

fading that he is finding a huge fat

of each diversity in this particular stamp all

and

you know trying to understand you know is that another way in which that is

is coming into you know the group or what's going on and so she's taking

into that i think you know it depends on the th adding you know the

degree to which differences in age will bring about those the fax

sometimes actually in more high tech settings it's not necessarily better to the older

in the it not necessarily the case that each diversity will automatically read the status

of a

but i do wanna definitely

you know i get

those something that i think with you know was saying about

you know readers

and what readers can do that can affect the

and i think definitely you know and some of the research again i'm at edmonton

is another person who's done some work on the

demonstrating that the way that the leader operate within the team like they're

there

appended feature

what different people think to invite their credit for them to admit when they don't

know things you know it's that arise seems to really make a difference in terms

of how much the status difference as

given the way of the group functioning so something that again you know

the median organisational culture around it but then there's something at the t mobile i

think they can be done trying to gated a little bit

and the is that

still

well once

also

in addition

you just

well the utterances

call this derivation

representation is integrated maybe

i guess of so

this became so literally research on

alright

so we haven't done study only think

well there is there is a study that we did this was with the

it was done in germany but with one of my collaborators we're they were looking

at collective intelligence and the groups where it was computer scientists and in other people

from other disciplines working on projects that were rated for

and i'm number of dimensions including innovation in we could find a strong

correlation between collective intelligence innovation

but also just you know theoretically

based on the very point two were making

other things that we know that we innovation and

learning

integration of expertise et cetera

coordination

our old things that lead innovation as well and so i think you know it's

it's just a small logical extension

that collective intelligence would

you can only do nothing so held

other than that and

impulse there working together or the

so we move or vision the you may or may be will in the u

to do you really uttered and when the more common well so how

you saw a copula can how can we will use the main issue

a role one syllable

all well

starting point

well known that is

and use them to you know what we do

how can we will

really

so but the and who wish

in fact another example that sometimes i hope and talks on the topic with comes

from support

and because of exactly words and you can have some very good players and the

team

you know

okay players great team you know what what's the different i would say that so

when a lot of our study

one that i was mentioning and some others

we measure collective intelligence pretty early on in the group slight if

and it's pretty predictable things that happen later

now we are not intervening to try to change the collective intelligence of the group

i mean that something that we're

you know trying to conduct some studies on vowel to see how that can be

done

but you know

what we know there's a whole you know kind a decade the literature i

group performance and norms and how

patterns get that very early in a group slight intend to be sustained like absent

some sort of intervention to change how that group is interacting

there are certain patterns that are sustained i mean in we viewed all the time

when i teach a class

people tend if anything c

the thing people talk all the time some people don't talk at all you know

and the atoms and like some sort of shaking things that rate as humans we

seem to be just wired a falling the patterns in follows patterns

fairly consistent rate

and so i would say that

in either you know the very do more diverse groups adding are in another group

standing often what makes the difference between a group that start slow and stays low

and one that gets better it's somebody something

trying to change there's patterns in get

people more bald as they get more comfortable

you know pull people out who are withdraw and you know kind of

quiet down the people who are dominating we know whatever the thing is that's going

and so you know i think

unlike individual intelligence we can't really change an individual's intelligence

sure of doing some sort of brain surgery

you know in collective intelligence you know i think there are lots of possibilities for

what we may be able to deal

and how

you see that

i he or

is more

at time

so we're actually right now in the middle of studying that questions that we spent

you know a couple years getting this web based tool designed that i mentioned earlier

and now we're in the middle this study where we you know where are trying

to measure collected collective intelligence and multiple point at multiple waiting time

the average we have so far suggests that at then an intervention it remains pretty

stable in our attributions are pretty much what i was just describing which is that

groups on the patterns and those patterns

you know that have than something shaking you know that kind of how they operate

and so you know things are fairly stable but we do still and again sometimes

the interventions there are fairly simple so my collaborator the one dimension sandy pentland humid

the social badges and then have also maybe things

which give groups realtime feedback about the distribution of their communication

and so they either have little

displays on the wall that are kind of indicate a you know who how much

each person is talking or they have there is a q one that i thought

where there is a ball there though the table is a display and there is

a ball in the centre and the group object is to

keep the ball in the centre but of one person a dominating the conversation like

right now the ball would flow for me

you know the then the with the kind of quiet down and let it you

know that make it a in the centre

and they find actually they look at you know performance that decision making kinda tasks

and they find that groups can work better when they get the speaker

about relative contribution

o two

i well

i k o k o nine or you know huh

but i

as i used it well i will use them as well

well you know i already know that users at last actually star

right you just see most

i mean you know i think that it's of my collaborator time alone at mit

whose you directs the centre for collective intelligence there

it's definitely interest in a nice idea that the organisation allowable

you know indefinitely the

you know possibilities or even you know multi organisational level where you know you can

setup collaborations a few well

that make you know the best use of what the individuals you know bring to

that collaboration you know even at that scale

in addition

all possible

which was a lot this is it's

one

i really with that but the researchers really controversy all

so i don't know the accuracies are baying i haven't you know i'm familiar with

the research in the book having read

you know that book

you know are repeated that work but

i would save to their number website now luminosity is one and you know there

are a number of others that both cognitive psychology researchers to

are interested in seeing if a can actually improve particular brain process the working memory

is often one of the key ones you know and

you know people's response speed you know how task they are attention is another you

know if we can improve this process it can we actually increased intelligence

and i would say at least studies that i'm familiar where

suggests that off and the

realm in which the effects

generalize is pretty limited

you know in so

on them a c d you can play these games it'll improve your memory your

attention and you're processing speed and generalized is to environments that are

you know those games are similar but it probably doesn't seem to generalize to you

know what you that do when you go back to your job and work with

your colleagues so

there's a lot of controversy

i would say that the most consistent added and comes from early intervention you know

and so that spend a lot of the support for head start programs and things

like that

in terms of enriching kids environments but

most of the evidence really suggest that those interventions have to happen early

so i o

in our is

well

where

you

but it is or so

so i

for me

if i

well as i say at all

where we're at her

so far

at a

it remember

dimensions

one is not sure that

i know there's huh

you haven't seen any of what content captured

okay

and with some

you are so i

unfortunately so you know there is a growing body of data there is a study

that came out and academic journal not too long ago looking at

after action reviews and there are facts for group learning and it was pretty convincing

evidence that it's definitely practise that is worth

doing for sure

if you're not only in terms of just you know

not making the mistakes you made here again but even you know

performance a novel domain so

you know just as a group we figure out things that we should just pay

attention to more broadly

so i would say that is definitely true i would also say that the organisation

there are few then i'm starting to get to know i don't think i can

name them yet but

consulting company where they also della turnover in people changing teams than that

it seems like

in and often differs something that

you know folks the total in a particular stress you can't really control but the

more broadly the information can be shared an integrated silence data

right in a document about this project

i am actually adding to expand the and involving document about projects of this kind

you know and that kind of creating more integrated knowledge than having a bunch of

individual reports where we you know here about how that project where and have to

do the integrating ourselves

but that involves quite a bit of

coordinated infrastructure and

openness of knowledge sharing for that to happen in a known a lot of organisations

that doesn't exist

but it works well and that kind in this consulting contracts that i studied

and it is in one

i guess the closest thing that a lot of people are familiar with is like

we could p d a right

so instead of

having documents that are in folders we key

and you know in several more working on projects of this time you know one

of our steps we you are team after action review and we all

work on expanding

the whiskey that is for projects of this kind

and it just sort of

you know how everybody is familiar with what other teams a put in there is

have to go there to have their own stuff

at that it added in this more integrated way

i met

it's to be used for

it is o g

o s

they are

i know

however i

yes

right

we wish

and

or it where i

there is there is no

you is

checked and

what about

huh

well i still

i will but also on line

as well

said issue

but

she

it is interesting there is used

you know

i was just

ryan example he's waves collaboration just remind me a growing number of collaboration for nine

have a you know involve a distributed collaborators which is why am i go to

meeting a lot but well have to take notes or do things in a double

dark as well so even note on our meeting you know in it

like okay so i'm gonna break down the you know we need is gonna do

x y z you know and so then by the end of the meeting you

know we have a set of collected notes that we can work you know work

off of so

i think that there is you know the technologies are enabling that's more so the

the quantities of static documents better in some folder have to go find that

you know or are starting to fade

okay thank you

right

well thank you i'm and enjoy talking with you about it and i hope it

with helpful or

you know that provoking for real and i look for the hearing more about what

you all the phi do

i m

i see i

okay

all right now mentioned it's not

this presentation

sure

between

okay alright makes

take care

here

Collective Intelligence and Team Learning (July 2014)

BBL

Dr. Anita Woolley, CMU