1
00:00:00,050 --> 00:00:03,830
Lesson managing cultural and behavioural change in AI teams.

2
00:00:03,830 --> 00:00:10,250
Managing cultural and behavioural change in AI teams is a multifaceted process that involves addressing

3
00:00:10,250 --> 00:00:15,260
both the organisational culture and the behaviours of the individuals within the team.

4
00:00:15,830 --> 00:00:21,710
This is essential for fostering an environment that supports ethical AI development and accountability.

5
00:00:22,460 --> 00:00:28,550
The complexity of AI projects, coupled with the ethical ramifications of AI deployment, necessitates

6
00:00:28,580 --> 00:00:31,250
a deliberate approach to managing these changes.

7
00:00:32,990 --> 00:00:39,830
AI teams are typically composed of diverse professionals, including data scientists, engineers, ethicists,

8
00:00:39,830 --> 00:00:45,170
and domain experts, each bringing unique perspectives and cultural backgrounds.

9
00:00:45,740 --> 00:00:48,170
This diversity can be a double edged sword.

10
00:00:48,170 --> 00:00:53,810
While it fosters innovation and comprehensive problem solving, it also presents challenges in aligning

11
00:00:53,840 --> 00:00:58,940
team members values and behaviours with the overarching goals of ethical AI governance.

12
00:00:59,790 --> 00:01:06,480
One critical aspect of managing cultural change is the establishment of a shared vision that encapsulates

13
00:01:06,480 --> 00:01:09,450
ethical principles and accountability standards.

14
00:01:09,870 --> 00:01:14,880
This vision should be clearly communicated and consistently reinforced by leadership.

15
00:01:15,780 --> 00:01:21,570
According to Shane, organizational culture is shaped by the values, beliefs and assumptions shared

16
00:01:21,570 --> 00:01:22,620
by its members.

17
00:01:22,950 --> 00:01:28,530
For AI teams, this means creating a culture where ethical considerations are integral to the decision

18
00:01:28,560 --> 00:01:31,380
making process, rather than an afterthought.

19
00:01:31,710 --> 00:01:37,050
Leaders must lead by example, demonstrating a commitment to ethical standards and encouraging open

20
00:01:37,050 --> 00:01:39,750
dialogue about potential ethical dilemmas.

21
00:01:41,370 --> 00:01:47,400
Behavioral change, on the other hand, involves modifying individual actions to align with the desired

22
00:01:47,400 --> 00:01:48,660
cultural shift.

23
00:01:49,170 --> 00:01:54,840
This can be achieved through targeted training programs, workshops, and continuous education initiatives

24
00:01:54,840 --> 00:01:58,500
that emphasize the importance of ethics in AI development.

25
00:01:59,120 --> 00:02:05,030
A study by Fog highlights that behavior change is more likely to occur when individuals are motivated,

26
00:02:05,060 --> 00:02:10,310
have the ability to perform the desired action, and are prompted by appropriate triggers.

27
00:02:11,000 --> 00:02:16,700
For AI teams, this might involve providing incentives for ethical behavior, ensuring team members

28
00:02:16,700 --> 00:02:22,520
have the necessary skills and knowledge, and creating an environment where ethical practices are regularly

29
00:02:22,520 --> 00:02:24,110
encouraged and rewarded.

30
00:02:26,240 --> 00:02:32,870
A practical example of managing both cultural and behavioral change in AI teams can be seen in the implementation

31
00:02:32,870 --> 00:02:34,610
of ethical review boards.

32
00:02:35,030 --> 00:02:41,000
These boards, composed of both internal and external members, serve as a checkpoint for AI projects,

33
00:02:41,000 --> 00:02:45,230
ensuring that they adhere to ethical guidelines and accountability standards.

34
00:02:45,680 --> 00:02:51,110
The presence of such boards sends a strong message about the organization's commitment to ethical AI,

35
00:02:51,110 --> 00:02:54,890
and creates a formal mechanism for addressing ethical concerns.

36
00:02:55,610 --> 00:03:01,240
Moreover, it encourages team members to consider the ethical implications of their work from the outset,

37
00:03:01,240 --> 00:03:05,260
fostering a proactive rather than reactive approach to ethics.

38
00:03:07,330 --> 00:03:13,720
Statistics from the AI Now Institute report underscore the importance of managing cultural and behavioural

39
00:03:13,720 --> 00:03:14,500
change.

40
00:03:15,100 --> 00:03:21,550
The report found that only 15% of AI practitioners felt their organisations were adequately addressing

41
00:03:21,550 --> 00:03:23,980
ethical concerns in AI development.

42
00:03:24,700 --> 00:03:28,570
This highlights a significant gap between the ideal and the reality.

43
00:03:28,600 --> 00:03:34,210
Emphasizing the need for a structured approach to managing these changes by embedding ethical principles

44
00:03:34,210 --> 00:03:39,970
into the organisational culture and promoting behaviours that align with these principles, organisations

45
00:03:39,970 --> 00:03:45,850
can bridge this gap and ensure that their AI initiatives are both innovative and responsible.

46
00:03:47,560 --> 00:03:51,970
Another critical element is the role of continuous feedback and improvement.

47
00:03:52,450 --> 00:03:58,300
AI teams should be encouraged to regularly reflect on their practices and seek feedback from peers,

48
00:03:58,300 --> 00:04:00,430
stakeholders and end users.

49
00:04:00,730 --> 00:04:04,750
According to Edmondson, creating a psychologically safe environment where.

50
00:04:04,780 --> 00:04:11,230
Team members feel comfortable sharing their concerns and suggestions is vital for continuous improvement.

51
00:04:11,800 --> 00:04:17,320
This approach not only helps identify potential ethical issues early on, but also fosters a culture

52
00:04:17,320 --> 00:04:19,390
of transparency and accountability.

53
00:04:20,980 --> 00:04:26,590
In addition to internal mechanisms, external collaboration and benchmarking can be valuable tools for

54
00:04:26,590 --> 00:04:29,080
managing cultural and behavioural change.

55
00:04:29,950 --> 00:04:35,740
Engaging with industry consortia, academic institutions and regulatory bodies can provide AI teams

56
00:04:35,740 --> 00:04:40,030
with insights into best practices and emerging standards in AI ethics.

57
00:04:40,540 --> 00:04:46,900
For instance, the partnership on AI, which includes members from academia, civil society and industry,

58
00:04:46,900 --> 00:04:51,880
offers a platform for sharing knowledge and developing guidelines for ethical AI.

59
00:04:52,120 --> 00:04:57,470
Such collaborations can help AI teams stay abreast of the latest developments and ensure that their

60
00:04:57,470 --> 00:05:00,230
practices align with broader industry standards.

61
00:05:02,300 --> 00:05:08,300
Furthermore, the integration of ethical considerations into the technical development process is crucial.

62
00:05:08,720 --> 00:05:14,180
This can be achieved through the adoption of ethical design frameworks and methodologies such as value

63
00:05:14,210 --> 00:05:15,440
sensitive design.

64
00:05:16,460 --> 00:05:22,160
VSD, as described by Friedman, Kohn, and Borning, involves identifying and addressing the values

65
00:05:22,160 --> 00:05:25,430
of stakeholders throughout the design and development process.

66
00:05:26,030 --> 00:05:32,180
For EI teams, this means incorporating ethical considerations into every stage of the AI life cycle,

67
00:05:32,180 --> 00:05:36,230
from data collection and model training to deployment and monitoring.

68
00:05:36,830 --> 00:05:41,720
By embedding ethics into the technical workflow, organizations can ensure that ethical principles are

69
00:05:41,720 --> 00:05:45,770
not just abstract ideals, but are concretely applied in practice.

70
00:05:48,380 --> 00:05:54,190
Managing cultural and behavioural change in EI teams also requires addressing potential Resistance to

71
00:05:54,220 --> 00:05:54,970
change.

72
00:05:54,970 --> 00:06:00,820
Resistance can stem from various sources, including fear of the unknown, perceived loss of autonomy,

73
00:06:00,820 --> 00:06:04,270
or skepticism about the benefits of ethical practices.

74
00:06:04,810 --> 00:06:10,540
Kotter suggests that successful change management involves creating a sense of urgency, building a

75
00:06:10,540 --> 00:06:15,910
coalition of change advocates, and communicating a clear vision for I teams.

76
00:06:15,910 --> 00:06:21,430
This might involve highlighting the risks of unethical AI, such as reputational damage and regulatory

77
00:06:21,430 --> 00:06:28,570
penalties, and showcasing the benefits of ethical AI, such as increased trust and long term sustainability.

78
00:06:30,280 --> 00:06:35,800
Lastly, it is essential to measure and evaluate the effectiveness of efforts to manage cultural and

79
00:06:35,800 --> 00:06:37,120
behavioural change.

80
00:06:38,140 --> 00:06:44,230
This can be done through regular assessments, surveys and audits that track progress and identify areas

81
00:06:44,230 --> 00:06:45,190
for improvement.

82
00:06:45,850 --> 00:06:51,400
Metrics might include the number of ethical training sessions completed, the frequency of ethical reviews

83
00:06:51,400 --> 00:06:55,730
conducted, and the level of employee engagement in ethical discussions.

84
00:06:56,240 --> 00:07:01,730
By systematically evaluating these efforts, organizations can ensure that their initiatives are making

85
00:07:01,730 --> 00:07:05,420
a tangible impact and continuously refine their strategies.

86
00:07:07,220 --> 00:07:13,730
In conclusion, managing cultural and behavioural change in AI teams is a complex but essential task

87
00:07:13,730 --> 00:07:16,970
for fostering ethical AI development and accountability.

88
00:07:17,420 --> 00:07:23,660
It requires a multifaceted approach that includes establishing a shared vision, promoting ethical behaviors,

89
00:07:23,660 --> 00:07:30,110
implementing formal mechanisms for ethical oversight, encouraging continuous feedback and collaboration,

90
00:07:30,110 --> 00:07:37,040
integrating ethics into the technical workflow, addressing resistance to change, and measuring progress.

91
00:07:37,040 --> 00:07:42,800
By adopting these strategies, organizations can create a culture where ethical considerations are ingrained

92
00:07:42,830 --> 00:07:49,190
in the fabric of AI development, ensuring that their AI initiatives are not only innovative, but also

93
00:07:49,190 --> 00:07:51,170
responsible and trustworthy.
