WEBVTT

1
00:00:00.293 --> 00:00:02.960
(upbeat music)

2
00:00:04.920 --> 00:00:06.600
<v ->Leaders face significant challenges</v>

3
00:00:06.600 --> 00:00:09.540
as they seek to capture the value from GenAI.

4
00:00:09.540 --> 00:00:11.250
Businesses need to rapidly deploy

5
00:00:11.250 --> 00:00:12.690
and scale the technology,

6
00:00:12.690 --> 00:00:14.430
while ensuring that it's proficient,

7
00:00:14.430 --> 00:00:16.088
safe and equitable, secure,

8
00:00:16.088 --> 00:00:19.140
and compliant with regulation and policy.

9
00:00:19.140 --> 00:00:21.600
Yet as we've engaged with countless clients

10
00:00:21.600 --> 00:00:24.243
across industries, we consistently see underinvestment

11
00:00:24.243 --> 00:00:27.330
and lack of capability to effectively test

12
00:00:27.330 --> 00:00:30.360
and evaluate GenAI systems at scale.

13
00:00:30.360 --> 00:00:33.600
And that can create some real challenges.

14
00:00:33.600 --> 00:00:36.450 line:15% 
Failing to identify risks early can mean guardrails

15
00:00:36.450 --> 00:00:38.100 line:15% 
aren't part of initial product design,

16
00:00:38.100 --> 00:00:39.630
leading to costly redevelopment

17
00:00:39.630 --> 00:00:41.880
late in the product development lifecycle.

18
00:00:41.880 --> 00:00:44.670 line:15% 
Not adequately planning resources and time

19
00:00:44.670 --> 00:00:47.640 line:15% 
for test and evaluation can lead to schedule delays

20
00:00:47.640 --> 00:00:49.560
and budget overruns, or worse,

21
00:00:49.560 --> 00:00:53.520
deploying product into production with unmitigated risks.

22
00:00:53.520 --> 00:00:56.730
All this can mean unintended harms for customers,

23
00:00:56.730 --> 00:01:00.810
and for businesses: increased exposure to financial,

24
00:01:00.810 --> 00:01:04.144
reputational, legal, and regulatory risks.

25
00:01:04.144 --> 00:01:06.811
(upbeat music)

26
00:01:09.660 --> 00:01:12.630
Responsible AI cannot be an afterthought.

27
00:01:12.630 --> 00:01:14.760
Test and evaluation needs to be incorporated

28
00:01:14.760 --> 00:01:16.740
throughout the product development lifecycle,

29
00:01:16.740 --> 00:01:18.360 line:15% 
from ideation to development,

30
00:01:18.360 --> 00:01:19.860 line:15% 
to post-deployment monitoring.

31
00:01:20.730 --> 00:01:24.480 line:15% 
Combining human-based and automated test and evaluation

32
00:01:24.480 --> 00:01:26.700
is the most effective and scalable way

33
00:01:26.700 --> 00:01:30.540
to ensure GenAI systems deliver value and mitigate risks.

34
00:01:30.540 --> 00:01:32.370 line:15% 
Educating the entire workforce

35
00:01:32.370 --> 00:01:34.230 line:15% 
on test and evaluation concepts

36
00:01:34.230 --> 00:01:36.120 line:15% 
builds awareness and understanding,

37
00:01:36.120 --> 00:01:38.760
while fostering a culture of shared responsibility

38
00:01:38.760 --> 00:01:42.603
to create GenAI systems that deliver, not destroy, value.