00:00:13
so this morning my alarm went off at
00:00:16
6:00 a.m. as soon as it ran I got up
00:00:20
really quickly I chose to go for a quick
00:00:22
run I decided on a healthy breakfast of
00:00:25
yogurt and fruit while listening to
00:00:27
classical music I chose to spend some
00:00:30
quality time with my children because I
00:00:33
would not see that much today I got
00:00:36
dressed with the outfit I had laid out
00:00:38
the night before got in the car and
00:00:41
decided to drive here while listening to
00:00:44
a mindfulness podcast to calm myself for
00:00:47
giving this talk I'm kidding of course
00:00:51
once the alarm went off in the unholy
00:00:54
hour of 6:00 a.m. I press the snooze
00:00:57
button until was no longer possible I
00:01:00
dragged myself out of bed in a foul mood
00:01:03
and then I texted my friend to complain
00:01:06
about it why do I have to do this
00:01:08
you know I debated then for quite a long
00:01:11
time was whether I should leave my hair
00:01:14
down or put it up I got ready very
00:01:17
quickly and then I decided to ask my
00:01:20
husband to drive her here so that I
00:01:22
could spend the ride looking at my notes
00:01:24
calling my mom and all these things so
00:01:27
what does this story tell us first that
00:01:31
I need to get a bit more organized
00:01:33
second that I like the rest of you make
00:01:37
decisions all the time I have been
00:01:41
talking for about two minutes now and
00:01:43
probably you think you have not made any
00:01:45
decisions since I started talking but
00:01:47
actually you have whether to scratch
00:01:50
your nose whether to respond to that
00:01:53
notification on your phone you have also
00:01:56
already probably decided whether my talk
00:01:59
will be interesting and whether you will
00:02:00
commit to it or whether you will start
00:02:03
playing with your phone talking to the
00:02:05
person next to you whom I hope you know
00:02:08
or just keep looking at the time until
00:02:11
the coffee break all day every day
00:02:15
whether you realize it or not you make a
00:02:19
lot of decisions our decision-making is
00:02:23
largely based on
00:02:24
things we have learned an experience in
00:02:26
the past on associations we have made in
00:02:30
our brains and sometimes they are even
00:02:33
based on the pros and cons balancing
00:02:36
them looking at alternatives calculating
00:02:39
probabilities and making a decision that
00:02:42
will maximize benefit in the long term
00:02:44
not so often though right
00:02:47
most often our decisions are based on
00:02:49
our impulses what we feel our mood our
00:02:53
biases sometimes even how much we have
00:02:56
had to drink right remember that text we
00:03:00
all have that text so the way we make
00:03:04
decisions it's about to fundamentally
00:03:07
change as we enter this new era of
00:03:10
technological advancements the era of
00:03:13
digital the era of big data one big
00:03:16
change that is about to come away is
00:03:19
around the way we humans will be making
00:03:22
decisions both small and big ones in the
00:03:27
era of big data decision-making will
00:03:31
move away from impulsive intuitive and
00:03:34
sometimes guided by drinking to
00:03:37
decisions based on data and evidence and
00:03:40
our partners for making all these
00:03:42
decisions will be intelligent machines
00:03:45
so firstly what is this big data I keep
00:03:48
talking about why is everyone talking
00:03:51
about it and how is it linked to
00:03:53
decision making big data describes large
00:03:57
and diverse sets of information that
00:04:00
basically keep growing because as you
00:04:04
very well know every aspect of our lives
00:04:06
has now been digitalized the way we work
00:04:09
they will communicate the way we
00:04:11
socialize even the way we fall in love
00:04:13
has been digitalized the word is now
00:04:16
full of data all activity that will
00:04:19
perform can now be locked and saved and
00:04:22
as a result the word now is packed with
00:04:25
an inconceivable amount of digital
00:04:28
records which we now call big data big
00:04:32
data is well big it's fast
00:04:36
and it comes in many different forms it
00:04:38
can be numbers it can be text can be
00:04:41
images that can be videos but what does
00:04:44
this have to do with decision making let
00:04:47
me share with you a story a bit more
00:04:50
than a decade ago two researchers at the
00:04:53
University of Cambridge in the UK
00:04:55
decided to investigate whether those
00:04:57
digital records of human behavior could
00:05:00
be used to predict someone's personality
00:05:03
what they wanted to see whether was
00:05:05
whether you could predict things like
00:05:07
openness and intelligence or even other
00:05:11
more intrusive things like sexual
00:05:14
orientation ethnic origin and political
00:05:17
views the digital records which they
00:05:20
would use would be simply people's
00:05:22
Facebook Likes couple of years later
00:05:26
they published a study where they showed
00:05:29
the results on data that they had
00:05:31
collected from 58,000 people who had
00:05:34
given them access to their Facebook
00:05:36
profiles and had also answered some
00:05:38
personality tests the results were
00:05:41
amazing they showed that on the basis of
00:05:45
only Facebook Likes the computer could
00:05:47
predict someone's skin colour with 95
00:05:50
percent accuracy and someone's gender
00:05:53
with 93 percent accuracy and what else
00:05:57
they did was that they showed what kind
00:05:59
of likes could be linked and could be
00:06:02
predictive of some personality traits
00:06:04
for example they showed that like that
00:06:08
were predictive of a high IQ where on
00:06:11
pages are just the lord of the rings'
00:06:13
science and kerry fries coincidentally
00:06:18
all things i like in contrast like that
00:06:24
could predict a low IQ where are you
00:06:29
worried
00:06:31
on pages related to makeup to
00:06:34
motorcycles and on I like being a Manny
00:06:37
groups I do like those even more
00:06:43
remarkably in a study the same
00:06:46
researchers published a couple of years
00:06:47
later what they did was that they asked
00:06:50
colleagues relatives and friends of the
00:06:53
participants to give their judgments of
00:06:56
the participants personalities and then
00:06:58
they compared them with the computers
00:07:00
judgments who do you think won
00:07:02
they showed remarkably that with only
00:07:06
ten likes the computer knew you better
00:07:09
and your colleague with only seventy
00:07:12
likes the computer you do better than
00:07:14
your friend with only a hundred and
00:07:16
fifty likes the computer you do better
00:07:18
than a relative and with 300 likes the
00:07:22
computer knew you better than your
00:07:23
partner the analysis they used were
00:07:27
really not that sophisticated and I hope
00:07:30
they're not watch yet they managed to
00:07:33
build a system that could make better
00:07:35
decisions than humans how did they
00:07:38
manage to do that
00:07:39
they managed it because they had big
00:07:41
data by having this big data set and
00:07:44
being able to analyze it they were able
00:07:47
to draw conclusions that were
00:07:49
statistically valid and to build an
00:07:51
intelligent decision making machine to
00:07:54
basically create artificial intelligence
00:07:58
the equation usually looks like this
00:08:00
data big data plus fancy or not so fancy
00:08:05
algorithms or months' equals artificial
00:08:09
intelligence the maps have been around
00:08:12
for a long time what we now have
00:08:14
available to us is big data and they are
00:08:17
already being used for effective our
00:08:20
decision-making Amazon is doing it
00:08:23
Amazon has been doing it for years they
00:08:25
have been shaping our shopping decisions
00:08:27
for years by giving us personalized
00:08:30
recommendations on products who might be
00:08:33
interested in buying using data of our
00:08:36
own past purchases and the past
00:08:39
purchases of people who are similar to
00:08:41
us Netflix is also doing it
00:08:44
for recommending a series or movies we
00:08:46
are likely to like both Amazon and
00:08:49
Netflix make these recommendations using
00:08:52
big data and artificial intelligence
00:08:54
methodologies making our everyday
00:08:57
decision-making much much easier so
00:09:00
christmas is coming up my 9 year old is
00:09:03
obsessed with Harry Potter so he's
00:09:05
getting a Harry Potter Lego set but
00:09:08
you're not supposed to tell so what else
00:09:11
can i buy him to go with it
00:09:13
do I need to worry about it no because
00:09:16
Amazon has filtered down all the choices
00:09:19
for me making that my decision-making
00:09:21
much much easier potential employers may
00:09:26
now be making decisions about us without
00:09:29
us knowing using our digital records if
00:09:33
Facebook Likes are so linked with
00:09:35
personality traits why not use digital
00:09:38
records like Facebook likes to see if I
00:09:41
will be resilient employee and a good
00:09:44
team player you could this photo come
00:09:47
back and bite me and yes I'm in the
00:09:50
haircut in the future
00:09:54
sometimes intelligent decisions maybe
00:09:58
even made before weaving knew that the
00:10:00
decision had to be made while I was
00:10:03
working as a researcher at the
00:10:05
University of Oxford part of my research
00:10:07
had to do with developing intelligent
00:10:10
tools to facilitate decision-making in
00:10:13
healthcare a project I was working on
00:10:16
had to do with use using a big data set
00:10:18
of health data from hospital patients in
00:10:22
order to build intelligent systems that
00:10:24
could predict whether something bad was
00:10:27
about to happen to someone while they
00:10:29
were in hospital how these systems
00:10:31
worked was by learning what combinations
00:10:35
of physiological characteristics could
00:10:37
signal future health crises and then
00:10:40
unalloyed was being generated now for
00:10:43
someone in hospital something bad
00:10:45
happening is not totally unexpected but
00:10:48
let's consider this scenario you're
00:10:50
peacefully sitting in your living room
00:10:52
watching a David Attenborough
00:10:54
documentary or Maria is today
00:10:57
you're watching the Kardashians so you
00:11:00
have been feeling a bit uneasy all day
00:11:03
but you are dismissing your symptoms
00:11:05
because you had a long day you don't
00:11:08
want to bother anyone you don't want to
00:11:10
make a fuss and suddenly you hear an
00:11:12
ambulance and there's a knock on the
00:11:14
door an intelligent system predicted
00:11:17
that you are about to have a heart
00:11:19
attack and they have come to take into
00:11:21
hospital making the right decision for
00:11:24
you
00:11:25
in those cases artificial intelligence
00:11:28
can literally proved to be a lifesaver
00:11:31
so as we enter this new era of digital
00:11:35
Big Data machine based decisions how
00:11:39
would my morning routine look like was
00:11:41
probably my sleep time in wake time
00:11:43
would have been optimized for me using
00:11:46
some fancy maths and the text message to
00:11:49
my friend could have been sent for me
00:11:51
using some prediction algorithm of what
00:11:54
I would be likely to want to tell my
00:11:56
friends my outfit would have been chosen
00:11:59
for me using a personalized algorithm
00:12:02
and the music I would listen to in the
00:12:05
car while coming here would have been
00:12:07
selected from me based on my
00:12:09
physiological signals as detected by my
00:12:13
car seat great huh actually the machine
00:12:16
could even give this talk for me in the
00:12:18
future but are we ready for all this are
00:12:22
we ready for data-driven and
00:12:24
evidence-based decision-making the
00:12:27
answer is not really because we humans
00:12:30
are not really built to be data-driven
00:12:33
decision makers technology is changing
00:12:37
our lives in a very fast pace but our
00:12:39
DNA has not really yet caught up
00:12:42
let's consider this scenario our
00:12:45
ancestors and sitting in the white
00:12:47
around the fire and they suddenly hear a
00:12:51
strange sound what could it be is it a
00:12:54
lion preparing for his evening meal
00:12:58
meaning us or is it some juicy prey that
00:13:02
could feed our family for a few days had
00:13:05
that been a world driven by artificial
00:13:08
intelligence our ancestors would have
00:13:10
pulled
00:13:11
their smartphones and would have checked
00:13:13
what the alcohol said that would have
00:13:15
made the decision according to the
00:13:17
algorithm with great accuracy in actual
00:13:22
fact our decisions had to make the
00:13:24
decision had to be made quickly with
00:13:26
little or no information at hand who do
00:13:30
you think survived the ones who stood up
00:13:32
and impulsively started running or the
00:13:35
ones who sticked around to make a
00:13:37
probabilistic assessment of the
00:13:39
situation let's have a show of hands who
00:13:42
thinks the impulsive ones who run away
00:13:44
survived okay who thinks the analysts
00:13:50
survived you're just trying to ruin my
00:13:54
point so most of you have common sense
00:13:59
and you've realized that it was the
00:14:03
impulsive ones who survived and that is
00:14:06
why we the descendants have evolved to
00:14:09
be largely impulsive decision makers
00:14:11
this impulsiveness along with many
00:14:14
biases we are carrying have plagued
00:14:17
human judgment and decision-making for
00:14:20
thousands of years as someone who works
00:14:24
on developing artificial intelligence
00:14:27
and on helping people and organizations
00:14:29
use it I can tell you for certain that
00:14:32
while those systems are not perfect in
00:14:35
many many situations machine based
00:14:38
decisions I excellent and they are
00:14:40
better than human decisions embracing
00:14:43
artificial intelligence can definitely
00:14:46
make our lives better if you want to
00:14:49
adapt to this new world and to benefit
00:14:52
from artificial intelligence in my
00:14:54
opinion there is two things you need to
00:14:57
accept the first is that for artificial
00:15:00
intelligence to work for you you have to
00:15:03
accept to expose yourself by sharing
00:15:06
your own personal data and secondly that
00:15:09
you will have to learn to trust
00:15:12
artificial intelligence even when it
00:15:14
goes against your own instincts and
00:15:17
impulses the recipe for success is not
00:15:21
allowing yourself to be passively swept
00:15:24
by this big day dine
00:15:26
artificial-intelligence wave with no
00:15:29
control of the situation
00:15:30
stay vigilant and educate yourselves
00:15:33
about which decisions you can trust at
00:15:36
the end of the day the decision about
00:15:39
whether you should follow the machines
00:15:42
decision should be yours and yours only
00:15:47
[Applause]
00:15:54
[Applause]
