1
00:00:00,050 --> 00:00:03,980
Lesson, harmonizing global AI laws and risk management frameworks.

2
00:00:03,980 --> 00:00:10,310
Harmonizing global AI laws and risk management frameworks is critical in ensuring that artificial intelligence

3
00:00:10,310 --> 00:00:15,890
technologies are developed and deployed in ways that are both ethical and beneficial to society.

4
00:00:16,760 --> 00:00:23,090
The diversity of AI applications, ranging from healthcare to finance, necessitates a robust and unified

5
00:00:23,090 --> 00:00:28,220
legal framework that can address the varied and complex risks associated with AI.

6
00:00:28,880 --> 00:00:31,760
This task requires international cooperation.

7
00:00:31,760 --> 00:00:37,610
Given that the borderless nature of AI technology means that regulatory inconsistencies can lead to

8
00:00:37,640 --> 00:00:39,170
significant challenges.

9
00:00:40,430 --> 00:00:46,700
The first challenge in harmonizing global AI laws is the disparity in regulatory approaches across different

10
00:00:46,700 --> 00:00:47,780
jurisdictions.

11
00:00:48,290 --> 00:00:54,050
For instance, the European Union has taken a proactive stance with the proposed AI act, which aims

12
00:00:54,050 --> 00:01:00,920
to establish a comprehensive legal framework for AI focusing on risk management, transparency and accountability.

13
00:01:01,400 --> 00:01:08,000
The EU's approach categorizes AI applications based on their risk levels from minimal risk to unacceptable

14
00:01:08,000 --> 00:01:11,420
risk and imposes corresponding legal requirements.

15
00:01:12,200 --> 00:01:17,810
On the other hand, the United States has adopted a more sectoral approach, with various federal agencies

16
00:01:17,810 --> 00:01:21,080
developing their guidelines tailored to specific industries.

17
00:01:21,110 --> 00:01:27,800
This fragmented approach can lead to regulatory gaps and inconsistencies, making it difficult for multinational

18
00:01:27,800 --> 00:01:30,710
companies to comply with diverse legal requirements.

19
00:01:33,200 --> 00:01:39,500
One effective strategy for harmonizing EI laws globally is through international cooperation and the

20
00:01:39,500 --> 00:01:42,020
establishment of multilateral agreements.

21
00:01:42,440 --> 00:01:47,990
Organizations such as the organization for Economic Cooperation and Development have been instrumental

22
00:01:47,990 --> 00:01:49,010
in this regard.

23
00:01:49,310 --> 00:01:55,580
The OECD's EI principles, endorsed by 42 countries, provide a foundation for national EI policies

24
00:01:55,580 --> 00:01:57,470
and international cooperation.

25
00:01:57,920 --> 00:02:03,980
These principles emphasize the need for AI systems to be robust, secure, and safe throughout their

26
00:02:03,980 --> 00:02:10,820
life cycle and for AI actors to be accountable for their operations by adhering to such internationally

27
00:02:10,820 --> 00:02:17,100
recognized Principles, countries can align their domestic regulations with global standards, facilitating

28
00:02:17,100 --> 00:02:19,740
greater consistency and cooperation.

29
00:02:20,880 --> 00:02:26,670
Risk management frameworks are integral to the regulation of AI, as they provide structured approaches

30
00:02:26,670 --> 00:02:31,950
to identify, assess, and mitigate risks associated with AI systems.

31
00:02:32,400 --> 00:02:37,920
Effective risk management frameworks should be dynamic, able to evolve with technological advancements,

32
00:02:37,920 --> 00:02:40,380
and adaptable to different contexts.

33
00:02:40,410 --> 00:02:47,610
The ISO, IEC 31,000 2018 Standard on Risk Management provides a comprehensive guideline that can be

34
00:02:47,610 --> 00:02:49,740
adapted to AI technologies.

35
00:02:49,770 --> 00:02:55,200
It emphasizes the importance of integrating risk management into organisational processes and decision

36
00:02:55,230 --> 00:03:00,810
making, thereby ensuring that risks are managed systematically across all levels.

37
00:03:02,280 --> 00:03:07,710
The use of risk management frameworks in AI also involves the implementation of ethical guidelines to

38
00:03:07,740 --> 00:03:11,730
address concerns related to bias, fairness and transparency.

39
00:03:12,210 --> 00:03:18,480
AI systems are often criticised for perpetuating existing biases, which can lead to unfair and discriminatory

40
00:03:18,480 --> 00:03:19,320
outcomes.

41
00:03:19,500 --> 00:03:26,520
For example, a study by Buolamwini and Gebru found significant biases in commercial AI gender classification

42
00:03:26,520 --> 00:03:32,550
systems, with error rates for darker skinned females being much higher than for lighter skinned males.

43
00:03:33,000 --> 00:03:38,040
These findings highlight the need for ethical risk management frameworks that incorporate measures to

44
00:03:38,070 --> 00:03:41,160
detect and mitigate biases in AI systems.

45
00:03:43,200 --> 00:03:47,550
Transparency is another critical aspect of risk management in AI.

46
00:03:48,030 --> 00:03:54,180
The black box nature of many AI algorithms makes it difficult to understand how decisions are made,

47
00:03:54,180 --> 00:03:56,520
which can lead to a lack of accountability.

48
00:03:56,970 --> 00:04:03,060
In response, there is growing advocacy for explainable AI, which aims to make AI systems more transparent

49
00:04:03,060 --> 00:04:04,140
and interpretable.

50
00:04:04,590 --> 00:04:10,020
An example of this is the General Data Protection Regulation in the EU, which includes provisions that

51
00:04:10,020 --> 00:04:15,630
require organizations to provide meaningful information about the logic involved in automated decision

52
00:04:15,660 --> 00:04:17,100
making processes.

53
00:04:17,880 --> 00:04:23,470
By incorporating transparency requirements into AI laws and risk management frameworks, Regulators

54
00:04:23,470 --> 00:04:29,200
can ensure that AI systems are accountable and that their decision making processes can be scrutinised.

55
00:04:31,960 --> 00:04:37,990
International collaboration is essential for the development and implementation of harmonised AI laws

56
00:04:38,020 --> 00:04:40,030
and risk management frameworks.

57
00:04:40,540 --> 00:04:46,870
Initiatives such as the Global Partnership on AI bring together experts from various countries to collaborate

58
00:04:46,870 --> 00:04:49,570
on shared challenges and opportunities in AI.

59
00:04:50,140 --> 00:04:56,770
Gpi's working groups focus on areas such as responsible AI, data governance and the future of work,

60
00:04:56,770 --> 00:05:01,240
providing a platform for knowledge exchange and the development of best practices.

61
00:05:01,960 --> 00:05:08,020
Such collaborative efforts are crucial in fostering a global consensus on AI governance and ensuring

62
00:05:08,020 --> 00:05:11,470
that regulatory approaches are aligned across borders.

63
00:05:12,670 --> 00:05:18,040
The role of standard setting bodies, such as the International Organization for standardization and

64
00:05:18,040 --> 00:05:24,580
the Institute of Electrical and Electronics Engineers is also pivotal in harmonizing AI laws and risk

65
00:05:24,580 --> 00:05:25,990
management frameworks.

66
00:05:26,650 --> 00:05:32,080
These organisations develop technical standards that provide detailed specifications and guidelines

67
00:05:32,080 --> 00:05:36,490
for the development, implementation and assessment of AI systems.

68
00:05:37,120 --> 00:05:43,330
For instance, the Ieee's Ethically Aligned Design document offers comprehensive guidelines for embedding

69
00:05:43,330 --> 00:05:47,230
ethical considerations into AI and autonomous systems.

70
00:05:47,890 --> 00:05:52,930
By adopting and adhering to these standards, countries can ensure that their regulatory frameworks

71
00:05:52,930 --> 00:05:59,680
are aligned with global best practices, thereby facilitating international cooperation and consistency.

72
00:06:00,850 --> 00:06:06,970
In conclusion, harmonizing global AI laws and risk management frameworks is essential for addressing

73
00:06:06,970 --> 00:06:11,890
the complex and multifaceted risks associated with AI technologies.

74
00:06:12,730 --> 00:06:18,130
This requires international cooperation, adherence to globally recognized principles and standards,

75
00:06:18,130 --> 00:06:22,540
and the implementation of dynamic and adaptable risk management frameworks.

76
00:06:23,080 --> 00:06:28,930
By fostering a cohesive and collaborative approach to AI governance, we can ensure that AI technologies

77
00:06:28,930 --> 00:06:35,110
are developed and deployed in ways that are ethical, transparent, and beneficial to society.
