Player is loading...
Round table
Panel
Saturday, Sept. 11, 2021 · 6:15 p.m. · 26m 59s
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:01
oh
00:00:03
oh
00:00:07
can you hear me
00:00:10
some going to have a first question for ms it's uh
00:00:14
then yeah regarding or that we've heard today you are a doctor
00:00:22
and we can see that we supported data we
00:00:25
are collecting now about are connected objects we or have
00:00:29
a connected objected as devices so this we serve
00:00:34
probably in the future provided that these data where uh
00:00:38
store and uh considered as medical data for predictive
00:00:42
medici and i would like to have your opinion
00:00:46
what are the gains or benefits and risks of these major uh data collect
00:00:51
and it are being collected by private industries also
00:00:59
yeah
00:01:04
yeah
00:01:09
they him yes
00:01:13
what's very interesting i think that you are asking the question
00:01:17
regarding the predictive missing or the use of uh uh in high
00:01:26
a i is being used or customised mitzi not
00:01:30
precision make it seem there are several uh names
00:01:34
and which uh uh is made of a sever elements we call it a four piece
00:01:41
personalise precision predictive and prevention
00:01:47
so where one hundred percent convinced and i don't need to be convinced by uh some
00:01:54
extent expectations and being convinced my some achievements today uh of this meeting is in there
00:02:01
here don't have not the production but use of the data that means
00:02:08
they're really easy we inning hacked up to collect or this data
00:02:14
and the use of i a. i. to classify and categorise and identify very
00:02:19
specific profiles of the four very everything
00:02:23
the pathology for those the quite actually yeah
00:02:30
for example or cancers uh which uh we
00:02:33
benefit from this approach or so the former co
00:02:36
genetics which allows it to heart better a quantification
00:02:42
of them made it seems and some orphan uh
00:02:45
and it's a it's a four hour in the past it took at
00:02:49
about twenty thirty e. is it to no more but then i within months
00:02:54
kind identify some chains now or these pathologies why it works quite where or
00:03:00
it's a very useful tool the reason why is that they
00:03:04
they gave us quite easy answers are quite easy information i mean
00:03:11
i talk about predicted made it's in now is that we want to
00:03:15
see that thanks to or did they die we will be able to predict
00:03:20
what's going to happen to you in ten easy we had i had it it or it's i'm air
00:03:25
article so you're going to believe in two hundred and fifty is all because maybe we'll achieve is where
00:03:32
in the production there is the use of the data yeah which to date
00:03:39
we we don't know how to combine and what i really liked in the first presentation by mister blah
00:03:45
is that there is no intelligence in a a artificial intelligence that today
00:03:51
we cannot say that when they would be able to
00:03:55
predict complex in it's it's such as the once we've
00:03:59
talked about we won't be able to predict if you're going to have a car accident or
00:04:04
or there is the major climate change or whatever so the
00:04:09
capacity of production is the most to we ski parts of
00:04:13
uh what we can for to personalise the meetings is the
00:04:16
most risky part under a scientific point of view and of course
00:04:21
this is also the most the whiskey element yeah under an economic point of view because
00:04:27
personally there are some not very unacceptable reactions as we see where there
00:04:32
you're not the vaccine i am vaccinated just to make sure things clear but when we say
00:04:38
that the people were we sure one not being vaccinated will not get access to the intensive care
00:04:44
is ethically this sees a how can we can we should do
00:04:48
we should tell they both smoker that they won't be able to be treated either
00:04:53
so are we then uh so to say predictive my there would be if uh
00:04:59
by you to a very and search and
00:05:03
element which is going to and then share quality of life and we're talking about the quality
00:05:08
of life if we say if we know that in ten years time maybe you we have to
00:05:15
you have to answer to to draft your your wheel
00:05:18
me it's not a win win situation in that case so
00:05:23
this is one issue the other one is the economic problem uh to
00:05:27
what insurance companies they which are going to say well now and i just
00:05:33
just to pinpoint that teaches companies they tried to do that
00:05:37
when there was the changing that being a about a human
00:05:41
humans you want it to you insert a but
00:05:45
uh this uh provision saying that each or genetic or data had been
00:05:49
asked by person had to be gave you as an information today insurance company
00:05:56
it's so what do you want to tell us no no no no it's very it's very interesting what you're tending
00:06:02
but i mean there is no intelligence in predictive made isn't that what you that's what you mean there is no intelligent
00:06:08
intelligence and not in productive meeting today there is no intelligence in pretty
00:06:14
maids in maybe i don't know what's gonna happen in twenty years but
00:06:17
just another point would go talking about quality of life and productive meeting to improve
00:06:23
quality of life to first think would be to ask people what they want at this
00:06:29
fighting e. discussions which we start eating
00:06:32
kinds engines with some projects but the
00:06:36
population at large i mean you for normal paid for or a g. then
00:06:43
say finally eventually we don't want to know we don't want to
00:06:48
know but this is opening the discussion about what we can know
00:06:54
and we shouldn't replace production by prevention now
00:06:59
we could make some prevention and we invite very little money for prevention so first money for
00:07:04
prevention and then production thank you very much i think this is a good message you are getting
00:07:11
so how could we win forethought transcend its prevention
00:07:15
with the date a we had data we have now so
00:07:18
maybe it opened it the bait and society to ask a question
00:07:22
two d. empowerment foundation and this is a late identity
00:07:31
uh huh nineteen year a friend of mine you need yet told me that the ethical issue
00:07:38
no i don't know what i'd like to know if what i do is good or not isn't
00:07:44
being addressed in a school is and the society's let addressing the
00:07:49
subject so where does it when she did come from the ethics
00:07:53
shoes should it be especially nice debate about the first of our our would like to say that we are living in a society where
00:08:01
there are a dividing everyone in a particular or a compartment department a we
00:08:07
have a low p. four engineer it's mm duct r. s. and so on
00:08:13
and there is the reality t. in there
00:08:17
have been digital world where or these partitions
00:08:22
do not exist anymore any longer we we have access to the no
00:08:27
lately can the four courses on line at e. p. f. a. r.
00:08:31
or in and uh you american uh university and while the tendency now is
00:08:37
to have not diverted course id but how we sent our society and we have
00:08:41
not only one uh a hat but several heights so
00:08:47
and when we have a fixed departments like a
00:08:52
different uh high schools i think that the uh
00:08:56
basically what we lack it's they're sure we basically narrate see
00:09:01
it's lacking and we should have an open the
00:09:05
gates not only having engineer is uh talking about to
00:09:10
i yeah i'd like to know how many years here are not scientists who is that
00:09:13
an engineer here or how good good so we are making uh who are working for discourse
00:09:22
so it shouldn't be a discussion only amount special rice beep
00:09:26
or and this is the reason why i think it is uh
00:09:31
really parents it to how i've uh more humanistic approach of society
00:09:36
as profitable i said we last this your mind is the approach
00:09:41
where we how historians uh uh the philosophers should participate in the d. b.
00:09:46
eight and not to have a vertical debate where we try to define or
00:09:51
several things that's why i appreciate it very much
00:09:53
discussions today because we talked about uh and fight
00:10:01
uh issues and i didn't fully against those uh easy see where we should do it just
00:10:07
to be framed it by big debate in the social meet yeah then even though the or something
00:10:13
why we read something easy switch air are we against our democratic feelings i think
00:10:20
these these that's a picture of society and we needed to accepted these and integrate it it
00:10:26
not to informing to buy us into the definition of these notions and
00:10:31
i we should open the bait and i really thank you for this uh
00:10:36
action your uh you've been uh doing for the last year
00:10:43
oh okay okay indeed i think this is the first to come
00:10:48
up between our foundations we can continue to do joint actions on
00:10:53
a social and societal issues i would like to hear much more
00:10:56
from you 'cause would've been for exciting to hear more from you
00:11:02
or maybe i could give the floor to uh some of our
00:11:07
old members uh of the telly molly foundation maybe each board member could
00:11:14
like give us a these conclusions of course i think we've covered a lot of ground
00:11:23
and yet there's still an awful lot of ground to cover so it's a difficult task
00:11:29
to troll conclusions i do is not necessarily to move forward to the
00:11:36
on the basis of certainties but we have to listen to all the arguments we've heard this afternoon
00:11:41
to move forward in small steps towards quality of life
00:11:49
we also engineers uh myself in particular and
00:11:54
engineers in particular tend to want to devise solutions
00:12:00
the solution could come from a a alright
00:12:06
the cost we've seen attempts with the g. d. p. ah in the u. s.
00:12:12
to engulf legal experts in the process of
00:12:16
escape price to some strange sure result result
00:12:20
this means that um today every time we want to go onto a side we have to
00:12:25
a click on our preferences but that's not really the right solution
00:12:29
because we have to control the quality in quality quality and quantity of data
00:12:36
we've um moved away from the data file west when the window
00:12:42
morsel rules or regulations no we've moved into a much more regulated area
00:12:47
in which the individual so protected to much more
00:12:53
so the actors uh no longer doing any old thing so we'll uh regulating mall but maybe that's not the right solution is
00:13:00
that the right solution is that we should be able to use
00:13:03
the data that we have to do so in a controlled manner
00:13:08
for madison as well but isn't has to be able
00:13:11
to apply cancer treatments to people who are capable of being
00:13:16
healed being cute because statistics being shown that term they are people
00:13:22
who want more sensitive to those treatments that would be something fundamentally useful
00:13:28
and uh when we have interests um if you uh um
00:13:32
when a a piece of information becomes available somewhere in the will to
00:13:36
which is related to interested would be great to get access to it
00:13:39
but this has to be done on the basis of a collaboration is it's not the whole gist of a group but um
00:13:49
we have to really appropriate position off all models
00:13:56
of course it's nice to be able to play with these models uh
00:13:59
asuncion yeah it's nice to be able to play with some models but um
00:14:03
the many ideas for a happens and uh projects which are
00:14:08
in the pipeline thank you see which i i would like to
00:14:17
re focus of our attention on the values
00:14:22
because whatever the approach we forget to high rock eyes the approach
00:14:28
as an education is at this uh makes me think of our tendency to cut
00:14:35
up the or a difficulty in to cut out the objectives and uh we no longer
00:14:42
after a while we no longer know where we are we no longer where we started from or where we're going where we are
00:14:50
so uh as a medical go i've often a hostess problem
00:14:55
then as the uh third notion which is that the proposition
00:15:00
uh thinking of the french speaking swiss a better coke
00:15:04
when talking about this a question of a fraction being sued
00:15:11
but it works so people who'd surrounded child with paper so's not to feel his or
00:15:16
her skin i think we have to feel people's skin have to get close to people
00:15:22
and we have to go back to the simple
00:15:24
human notions like communication simplification uh respect and dialogue
00:15:32
but uh i have great difficulty in reconciling these two dimensions
00:15:38
of this description that i've just given a may seem uh
00:15:41
one dimension machines with the colours what i've just
00:15:45
described buttons the dimension of science experiments and so on
00:15:49
at the end of the day i have the impression that we no longer know
00:15:52
exactly where we started from where we're heading where we find ourselves at the moment
00:15:59
you've mentioned uh into discipline energy and multi disciplinary
00:16:03
to us but i think we have to but um
00:16:08
the people to uh a few little people them to to feel
00:16:13
to sense but uh what they want and to know what they need in order to succeed
00:16:18
in to a um a a them pretty happy that's what which i hardly dare use no
00:16:28
we have to introduce more humanism intro relations these are not
00:16:31
scientific notions of concepts of course offence to this but for me
00:16:37
they are pretty fundamental notions so science should come back to the service of this
00:16:43
fundamental notions i'm afraid that ah in yeah and so it's maybe working just for itself
00:16:54
soul problem
00:16:56
in a economics the economy we've seen that uh some
00:17:00
companies so structuring so much and the planning so much
00:17:04
that they uh uh really a living only for themselves the for forgotten
00:17:09
that they have to uh to exist in order to produce the uh
00:17:14
system is turning a guy in a circle going round in a circle i think he
00:17:21
some way a vision of life is that life is a complicated but maybe at the end of the day it's not uh
00:17:29
it's complicated as it may may seem if we go back to fundamental notions
00:17:34
so i think the more complicated the notions we developed the more
00:17:37
complicated things will seem thank you for your free commonsense based uh
00:17:44
approach shot commonsense which we saw at the at the beginning
00:17:49
yeah well i could summarise simply by saying that i agree with everything that's been said
00:17:54
uh but uh each and every idea could be pushed to the extremes of how
00:18:00
could i start well first of all i'm also an engineering problem paying attention yeah
00:18:06
and uh i'm a mound of the right
00:18:12
and the proud of it but you can be an engineer and be on the right of the political spectrum
00:18:18
and at the same time belief in a quality of life and humanism and other such qualities
00:18:23
but as an engineer i have to be very pragmatic so when talking about some things i i'd like to know what
00:18:29
we're talking about the present all the short term future all the long term future when talking about the quality of life
00:18:38
i think that's a rather idealistic utopian idea because uh each
00:18:43
person will come up with a different definition of quality of life
00:18:46
although of course it's a good thing to ask the question the same applies to ethics we know when someone is not ethical
00:18:54
but to say to someone really is ethical it's not possible is is uh like you know
00:18:58
same us trying to define um scalded life when we asked a question about existing systems
00:19:04
for so ross also to something interesting uh in this regard
00:19:10
but um in my presentation at the bottom what i
00:19:13
was trying to say was that most developments in a high
00:19:17
make us believe in things which don't exist told me models with one name
00:19:24
two maxima lies the number of kicks as what
00:19:27
mister stewart certain to maximise financial income and so on
00:19:32
and uh that is the fault of the present system uh even though i'm on the right of the political spectrum
00:19:39
to uh go out to make people believe certain things at all costs uh
00:19:44
in a bid to make profits short term profits on the backs of uh people who are suffering
00:19:49
uh patients for example in in the field of medicine this has
00:19:53
to be expressed much more clearly in media you've spoken about productive madison
00:20:01
uh yeah prefer to talk about computer assisted prevention doctors
00:20:05
to prevention insurance companies also do prevention if you're uh
00:20:08
a smoker your life insurance will costume all all health insurance even
00:20:14
uh when you reached fifty um your additional a complementary
00:20:18
health issues will cost you more fusion uh been a smoker
00:20:21
but to being fifty years old or what i think is shameful is the to conventional medicine i can um uh
00:20:29
refer you to some articles um which say that some of the engine is
00:20:36
uh uh say that um we should stop training uh radiologists because deep learning
00:20:42
does the job patent humans uh does the job but uh but um radiologist
00:20:48
uh you need about fifteen years of training to become a radiologist
00:20:52
it's clear that i'm a google cannot just train people in a few months to become a radiologist
00:21:03
and uh um the the these computer assisted tools computer based
00:21:07
tool so just to assist radiologist up to replace them with
00:21:14
no um if if you come along with the predictive for diagnostic system
00:21:17
before being able to use it the system has to be approved them
00:21:21
by a certain uh uh ministrations searches the few fruit mantra come instruction
00:21:27
in the u. s. it normally takes about uh ten years to get f.
00:21:31
d. a. approval in us there's only one which uses a little bit of a
00:21:38
a a high uh um to help radiologist thoughts um to detect
00:21:44
breast cancer and this has been the case since two thousand and two
00:21:49
uh so in certain cases uh um we can been equivalents uh between radiologists and uh
00:21:58
it i
00:22:00
but uh twenty years later only three percent of women accept to diagnosis by a
00:22:07
a computer rather than by a doctor because they prefer to speak to a real doctor
00:22:15
so you have to look at the human criteria all the machines so we're looking
00:22:20
at today based on financial criteria short term profit you have to be aware of that
00:22:26
and the objectives of the machines are optimised to make money the criteria to use to train
00:22:33
the machines that sebastian we're talking about art optimised things which make money as quickly as possible
00:22:39
at the expense of the insurance companies um the that expensive of people of individuals
00:22:50
and that this brings me back to the aims of our foundation
00:22:55
our aim is to maximise the impact on
00:22:59
the individual so this is a person related individual
00:23:03
try to rim related to the call to fly for let's roll the utopian because
00:23:07
it will not be possible to to draw up an exhaustive list of criteria that define
00:23:12
quality of life but come back to you would you uh uh accept
00:23:16
to be um we uh the object
00:23:20
of um a i assisted to prevention
00:23:27
even when talking about prevention um
00:23:32
of course it would be a interesting to know uh what other genetic factors which make
00:23:38
us more a problem to certain a a
00:23:40
certain uh diseases and that um constant cost um
00:23:47
it should be possible to do this with the the machine i just referred to so used by a three percent of um
00:23:55
three percent of 'em patients but the machine
00:23:59
uh the cost one thousand times morton radiologist
00:24:05
we have to create a climate of confidence among the general population it's something we've lost because of all these uh
00:24:11
i've uh tossing pasted tools and because of this uh falsified indicators
00:24:16
the aim being to make short term profit that's all there is i'm a a recording engineer
00:24:27
so what you say is we need more interest will confident so
00:24:32
that really is the key just under the moment everything is being done to break this a trust
00:24:38
and confidence i think there is one question from the floor will take this question from the floor
00:24:45
thank you very much of a question from the floor my the question was infected comments not to question
00:24:53
but all the things you do know presenting them of this often think we
00:24:58
if we all want a better quality of life apart from artificial intelligence which
00:25:05
would enable us to um reduce or eliminate to all these menial tasks
00:25:11
but uh would make things so much easier when talking about prevention
00:25:17
what i think is capital today is that we have to open up this uh to the to the public
00:25:23
we need to revolution in that respect and to what we
00:25:26
need is a to have a proper understanding of our use says
00:25:32
what i've been able to see it all the symposium some conferences and meetings that
00:25:37
deal with the subject i want always comes back to the problem of useful usage
00:25:43
and if we don't ask these questions today uh where are we going to be
00:25:48
fighting when it comes to interaction with
00:25:51
the between technologies and the human aspect to
00:25:56
google knows exactly where they want to head but i think that's a question we
00:26:01
want to we need to ask ourselves now if we want a better life style
00:26:05
i have nothing further to uh the says orally that really was uh the conclusion
00:26:11
thank you for big thank you to all our speakers supporting participated today
00:26:17
and thank you for older people who for helped
00:26:20
us with the uh technology the interpreters and other subcontractors
00:26:27
to have enabled us to bring about this interface
00:26:31
and who made possible these uh presentations in english
00:26:35
and french thank you mary for having organised event together with myself i yeah which you very nice evening
00:26:45
and uh
00:26:48
the uh pretty full be just outside the so we really started out of uh
Conference Program
Welcome words
Aurélie Rosemberg, Fondation Dalle Molle
Sept. 11, 2021 · 4 p.m.
Artificial intelligence and quality of life
H. Bourlard, Idiap Research Institute
Sept. 11, 2021 · 4:30 p.m.
Artificial intelligence to think like humans
Melanie Mitchell, Professor at the Santa Fe Institute
Sept. 11, 2021 · 4:45 p.m.
Towards human-centered robotics
Sylvain Calinon, Research Director at the Idiap Research Institute
Sept. 11, 2021 · 5 p.m.
Supporting sustainable transitions around the world through water technology
Eric Valette, Director of AQUA4D
Sept. 11, 2021 · 5:15 p.m.
Biometric security
Sébastien Marcel, Research Director at the Idiap Research Institute
Sept. 11, 2021 · 5:30 p.m.
Compatibility with humans: AI and the problem of control
Stuart russel, Professor of Computer Science and Smith-Zadeh Professor of Engineering, University of California, Honorary Fellow of Berkeley and Wadham College at Oxford
Sept. 11, 2021 · 5:45 p.m.
Model subjectivity at the heart of consciousness to make robots more human
David Rudrauf, Associate professor at the University of Geneva, Director of the laboratory of the multimodal modeling of Emotion and Feeling
Sept. 11, 2021 · 6 p.m.
Round table
Panel
Sept. 11, 2021 · 6:15 p.m.
Recommended talks
Mot de bienvenue
Aurélie Rosemberg, Fondation Dalle Molle
Sept. 11, 2021 · 4 p.m.
379 views