WEBVTT

92cc86fa-5930-4d61-9d00-fd1af9eacf9c-0
00:00:00.400 --> 00:00:01.280
Ben, welcome.

08b0cd69-6388-40a7-b36a-eee940ca681f-0
00:00:01.400 --> 00:00:02.160
How is GenAI

5777506e-484e-43bb-a3fa-2d69a0e82637-0
00:00:02.200 --> 00:00:04.400
impacting the role of insight
professionals?

66559f1d-99de-406d-bd18-5c121f8481f2-0
00:00:04.880 --> 00:00:08.722
Well, it's it's like another
tool, and it's letting you do

66559f1d-99de-406d-bd18-5c121f8481f2-1
00:00:08.722 --> 00:00:09.960
things much faster.

411ff8d5-9333-431d-bfe5-6fec2ba00b0f-0
00:00:09.960 --> 00:00:12.160
It's letting you be more
creative.

a3b621ca-2c7b-48f7-ae9e-0a804fb093ab-0
00:00:13.120 --> 00:00:16.211
It's giving you new
opportunities that simply didn't

a3b621ca-2c7b-48f7-ae9e-0a804fb093ab-1
00:00:16.211 --> 00:00:19.710
exist before because you can
mine huge amounts of data in a

a3b621ca-2c7b-48f7-ae9e-0a804fb093ab-2
00:00:19.710 --> 00:00:23.093
few seconds in a way that was
just not possible before it

a3b621ca-2c7b-48f7-ae9e-0a804fb093ab-3
00:00:23.093 --> 00:00:23.560
arrived.

f5adfaad-8d3e-4a97-b57a-28c6f045a526-0
00:00:24.240 --> 00:00:25.440
What is your approach to GenAI

2de8565e-6f32-4ffa-9404-3ed296be7595-0
00:00:25.440 --> 00:00:26.120
at Ipsos?

b84426ec-f852-4b20-ba8a-b56acae6dcc3-0
00:00:26.160 --> 00:00:30.132
So what we're doing is we see it
as a combination of human

b84426ec-f852-4b20-ba8a-b56acae6dcc3-1
00:00:30.132 --> 00:00:32.960
intelligence and artificial
intelligence.

7ed6669e-e5b7-4de6-ba31-0109bf90d8cf-0
00:00:33.200 --> 00:00:35.941
And so we're working on
improving our speed and

7ed6669e-e5b7-4de6-ba31-0109bf90d8cf-1
00:00:35.941 --> 00:00:39.540
productivity overall, but we're
also building new products and

7ed6669e-e5b7-4de6-ba31-0109bf90d8cf-2
00:00:39.540 --> 00:00:43.138
services for our clients, like,
you know, Explorer, which lets

7ed6669e-e5b7-4de6-ba31-0109bf90d8cf-3
00:00:43.138 --> 00:00:46.279
you look at innovations in a
completely different way.

225691fb-9267-419b-b8e8-9ece4132fa56-0
00:00:46.560 --> 00:00:49.883
So it's both more creativity,
but also just getting things

225691fb-9267-419b-b8e8-9ece4132fa56-1
00:00:49.883 --> 00:00:50.560
done faster.

8472ca7d-be55-4a2a-9cab-d9455e250ee6-0
00:00:50.800 --> 00:00:51.280
Interesting 

64ac9c70-f8ec-41c6-b68d-6a8310543751-0
00:00:51.280 --> 00:00:54.176
you're saying still human
intelligence and lots of

64ac9c70-f8ec-41c6-b68d-6a8310543751-1
00:00:54.176 --> 00:00:57.640
creativity, some things that
people don't normally associate

64ac9c70-f8ec-41c6-b68d-6a8310543751-2
00:00:57.640 --> 00:00:58.720
when we talk about 

f8ee21bd-7cec-4f42-b5c8-ee5af1bb1e0e-0
00:00:58.760 --> 00:01:00.520
GenAI and AI and technological
advance?

332941e8-2c99-482b-b62a-2f746ba295b0-0
00:01:00.520 --> 00:01:03.525
Well, I think there is a risk of
sort of dulling down to the

332941e8-2c99-482b-b62a-2f746ba295b0-1
00:01:03.525 --> 00:01:03.920
average.

3f5600d4-2e47-4fbd-833e-f5e0ca5ff4b7-0
00:01:03.920 --> 00:01:05.800
And you know, there are some
signs,

ecdc008d-0bed-4c70-a92b-b6dacbf738db-0
00:01:05.800 --> 00:01:08.524
for example, I mean, we're here
in Cannes and looking at

ecdc008d-0bed-4c70-a92b-b6dacbf738db-1
00:01:08.524 --> 00:01:09.720
advertising, for example.

63c1c35b-d5a8-49c1-be13-a50fdb61097f-0
00:01:09.720 --> 00:01:12.681
One thing that we can see is
that over the last six or seven

63c1c35b-d5a8-49c1-be13-a50fdb61097f-1
00:01:12.681 --> 00:01:15.838
years, there's less and less use
of humour in advertising, which

63c1c35b-d5a8-49c1-be13-a50fdb61097f-2
00:01:15.838 --> 00:01:18.994
is a shame because actually it's
one of the things that makes it

63c1c35b-d5a8-49c1-be13-a50fdb61097f-3
00:01:18.994 --> 00:01:19.479
memorable.

741f1137-1684-4006-8e47-d5ab714a41f9-0
00:01:19.840 --> 00:01:23.541
And actually what we found is
that AI isn't very good at

741f1137-1684-4006-8e47-d5ab714a41f9-1
00:01:23.541 --> 00:01:25.360
spotting people doing humor.

1225819b-981d-48b1-878c-dcbfb6427521-0
00:01:25.360 --> 00:01:28.775
It can sort of tell that
something's entertaining, but it

1225819b-981d-48b1-878c-dcbfb6427521-1
00:01:28.775 --> 00:01:32.485
can't really tell how good, how
funny something might be, yet.

1225819b-981d-48b1-878c-dcbfb6427521-2
00:01:32.485 --> 00:01:33.840
Yet. Yet indeed, quite.

20b86964-f84b-48d1-b57b-e7730a399166-0
00:01:34.000 --> 00:01:35.360
What are you most excited for
then?

b9cfc7a6-10e3-40d9-a203-669810d76122-0
00:01:35.520 --> 00:01:37.480
Talking about the future for AI,
GenAI

6bca7a43-46ed-4abc-85ab-7fafc387097a-0
00:01:37.480 --> 00:01:40.721
at Ipsos? Well, I think it's, I
mean, it's generally just

6bca7a43-46ed-4abc-85ab-7fafc387097a-1
00:01:40.721 --> 00:01:43.515
improving as a financial
company, as a commercial

6bca7a43-46ed-4abc-85ab-7fafc387097a-2
00:01:43.515 --> 00:01:46.421
business, it's generally
improving productivity and

6bca7a43-46ed-4abc-85ab-7fafc387097a-3
00:01:46.421 --> 00:01:49.941
letting people get things done
faster, taking the slog away to

6bca7a43-46ed-4abc-85ab-7fafc387097a-4
00:01:49.941 --> 00:01:52.680
let them spend more time
thinking about clients.

053a7288-1428-41f0-8478-f8cacb621e3c-0
00:01:53.000 --> 00:01:56.160
But I think we can see a future
where actually we start to see

053a7288-1428-41f0-8478-f8cacb621e3c-1
00:01:56.160 --> 00:01:58.920
the use of synthetic data, and
that's really exciting.

dbcedb23-0a39-46e6-a6bd-b0352dad4c86-0
00:01:59.160 --> 00:02:02.400
But you need to be very careful
about that, and smoke, but not

dbcedb23-0a39-46e6-a6bd-b0352dad4c86-1
00:02:02.400 --> 00:02:03.120
always inhale.

26c3b2c2-b3f4-4d22-a6a6-9e6b379c1364-0
00:02:03.720 --> 00:02:04.960
How—interesting.

060ead27-7ada-4f54-b6cc-588dcad40395-0
00:02:05.280 --> 00:02:05.840
How are you, 

0c8f280d-42da-4ed3-9085-725672fe81c1-0
00:02:05.920 --> 00:02:08.513
how are you making sure that you
don't fall into the potential

0c8f280d-42da-4ed3-9085-725672fe81c1-1
00:02:08.513 --> 00:02:08.760
traps?

bebd804b-9add-412e-b110-8700a38e0765-0
00:02:08.760 --> 00:02:11.080
Well, we use three principles.

1be0712e-c342-476a-bdc4-2fa7ea2edc09-0
00:02:11.080 --> 00:02:15.960
So, you know, we, we want truth,
transparency, and justice.

55fef663-2218-4c15-903d-cb8c203ff457-0
00:02:16.240 --> 00:02:19.972
So in other words, if you take
justice, is it just reinforcing

55fef663-2218-4c15-903d-cb8c203ff457-1
00:02:19.972 --> 00:02:20.920
existing biases?

fdd716b6-5956-412e-ad1c-738d86dba22b-0
00:02:20.920 --> 00:02:24.069
You know, if you ask AI and many
of the models now to tell you,

fdd716b6-5956-412e-ad1c-738d86dba22b-1
00:02:24.069 --> 00:02:27.268
you know, what a chief executive
looks like, it's probably going

fdd716b6-5956-412e-ad1c-738d86dba22b-2
00:02:27.268 --> 00:02:29.680
to look more like me than like
you. For example,

41baded7-402a-4730-9e7f-a26a3872ebb8-0
00:02:30.400 --> 00:02:31.280
is it transparent?

3f6c2bf1-4662-4591-8bf2-6eb9d8df7cc2-0
00:02:31.280 --> 00:02:34.763
We have to know how it works,
because if it's just black box,

3f6c2bf1-4662-4591-8bf2-6eb9d8df7cc2-1
00:02:34.763 --> 00:02:36.280
then there's some problems.

e0777d2e-dd49-4a48-9b71-352cd4d90d72-0
00:02:36.280 --> 00:02:38.520
So we need to know exactly how
it's working.

be898f6a-f989-47a7-859f-cdfec92030d3-0
00:02:38.840 --> 00:02:41.986
And finally, of course, we need
to make sure that it's not

be898f6a-f989-47a7-859f-cdfec92030d3-1
00:02:41.986 --> 00:02:44.440
hallucinating, which again, can
be a problem.

066bb421-630f-4c1a-a0b6-91537d48c400-0
00:02:44.960 --> 00:02:46.400
Ben, thank you so much for your
time. Pleasure.