1
00:00:00,050 --> 00:00:00,530
Lesson.

2
00:00:00,530 --> 00:00:03,350
Comparison of global ethical guidelines for AI.

3
00:00:03,380 --> 00:00:09,050
The comparison of global ethical guidelines for artificial intelligence is a pivotal discourse in the

4
00:00:09,050 --> 00:00:12,860
realm of responsible AI principles and trustworthy AI.

5
00:00:13,220 --> 00:00:19,070
Ethical guidelines for AI are essential, as they provide a framework for ensuring that AI technologies

6
00:00:19,070 --> 00:00:25,370
are developed and deployed in a manner that aligns with societal values, human rights, and legal norms.

7
00:00:25,610 --> 00:00:32,120
These guidelines also help mitigate risks associated with AI, such as biases, privacy violations,

8
00:00:32,120 --> 00:00:34,010
and unintended consequences.

9
00:00:34,520 --> 00:00:39,830
Various countries and organizations have developed their own ethical guidelines for AI, reflecting

10
00:00:39,830 --> 00:00:43,070
their unique cultural, social, and legal contexts.

11
00:00:43,430 --> 00:00:49,700
This lesson examines and contrasts some of the most influential global ethical guidelines for AI, highlighting

12
00:00:49,700 --> 00:00:53,000
their commonalities, differences, and implications.

13
00:00:54,350 --> 00:01:00,560
One of the most prominent ethical guidelines for AI is the European Union's ethics guidelines for trustworthy

14
00:01:00,590 --> 00:01:06,180
AI, Eye, published by the High Level Expert Group on Artificial Intelligence in 2019.

15
00:01:06,600 --> 00:01:10,800
The EU guidelines emphasize seven key requirements for trustworthy AI.

16
00:01:10,830 --> 00:01:12,480
Human agency and oversight.

17
00:01:12,510 --> 00:01:14,640
Technical robustness and safety.

18
00:01:14,670 --> 00:01:16,380
Privacy and data governance.

19
00:01:16,410 --> 00:01:20,610
Transparency, diversity, non-discrimination and fairness.

20
00:01:20,610 --> 00:01:24,210
Societal and environmental well-being and accountability.

21
00:01:24,210 --> 00:01:30,000
These principles are designed to ensure that AI systems are lawful, ethical, and robust across their

22
00:01:30,000 --> 00:01:31,230
entire life cycle.

23
00:01:31,800 --> 00:01:37,380
For instance, the principle of human agency and oversight ensures that AI systems do not undermine

24
00:01:37,380 --> 00:01:41,250
human autonomy and that there is always an option for human intervention.

25
00:01:43,290 --> 00:01:48,210
Contrastingly, the United States has taken a more decentralized approach to AI ethics.

26
00:01:48,240 --> 00:01:54,090
The US approach is characterized by sector specific guidelines rather than a unified framework.

27
00:01:54,870 --> 00:02:00,210
For example, the National Institute of Standards and Technology released a draft framework for managing

28
00:02:00,240 --> 00:02:06,950
AI risks in 2021, which focuses on accuracy, reliability and security of AI systems.

29
00:02:07,490 --> 00:02:12,890
The NIST framework aims to foster innovation while addressing the risks associated with AI.

30
00:02:12,920 --> 00:02:18,110
Recommending voluntary standards to ensure the trustworthiness of AI technologies.

31
00:02:18,500 --> 00:02:24,740
This approach reflects the US emphasis on innovation and market driven solutions, contrasting with

32
00:02:24,740 --> 00:02:26,930
the EU's more regulatory stance.

33
00:02:27,680 --> 00:02:34,370
China's Ethical Guidelines for AI, published by the Beijing Academy of Artificial Intelligence in 2019,

34
00:02:34,370 --> 00:02:37,790
reflect the country's unique socio political context.

35
00:02:38,330 --> 00:02:43,820
The Beijing AI principles emphasize the alignment of AI development with the nation's strategic goals,

36
00:02:43,820 --> 00:02:46,670
societal stability, and ethical values.

37
00:02:47,000 --> 00:02:52,820
These principles include harmony and friendliness, fairness and justice, inclusivity and sharing,

38
00:02:53,090 --> 00:02:56,480
respect for privacy and security, and controllability.

39
00:02:56,510 --> 00:03:03,120
A key aspect of China's guidelines is the emphasis on the collective good over individual rights, reflecting

40
00:03:03,120 --> 00:03:05,070
the country's collectivist culture.

41
00:03:05,100 --> 00:03:11,220
The principles also stress the importance of AI contributing to national security and social stability,

42
00:03:11,250 --> 00:03:15,240
highlighting the government's role in guiding and regulating AI development.

43
00:03:17,880 --> 00:03:23,370
Japan's approach to AI ethics, outlined in the Social Principles of Human Centric AI, published by

44
00:03:23,370 --> 00:03:29,400
the Japanese government in 2019, emphasizes the harmonious coexistence of humans and AI.

45
00:03:29,850 --> 00:03:35,520
The Japanese guidelines focus on five principles human centricity, education and literacy.

46
00:03:35,550 --> 00:03:36,780
Privacy protection.

47
00:03:36,780 --> 00:03:39,060
Security, and fair competition.

48
00:03:39,450 --> 00:03:45,180
The human centric approach ensures that AI technologies serve human well-being and societal good.

49
00:03:45,210 --> 00:03:51,150
Japan's guidelines also highlight the importance of public understanding and education about AI, indicating

50
00:03:51,150 --> 00:03:56,370
a proactive stance in preparing society for the integration of AI technologies.

51
00:03:56,820 --> 00:04:01,960
This approach aligns with Japan's cultural values of harmony and respect for human dignity.

52
00:04:03,760 --> 00:04:09,190
International organizations have also contributed to the development of ethical guidelines for AI.

53
00:04:09,460 --> 00:04:15,280
The organisation for Economic Co-operation and Development published its AI principles in 2019, which

54
00:04:15,280 --> 00:04:17,860
have been adopted by over 40 countries.

55
00:04:17,890 --> 00:04:25,480
The OECD principles emphasize inclusive growth, sustainable development, human centered values, transparency,

56
00:04:25,480 --> 00:04:29,290
robustness, security, safety and accountability.

57
00:04:29,320 --> 00:04:34,990
A notable aspect of the OECD guidelines is their emphasis on international cooperation and the need

58
00:04:34,990 --> 00:04:41,230
for global alignment in AI ethics, reflecting the global nature of AI technologies and their impact.

59
00:04:42,880 --> 00:04:48,010
Despite the differences in these guidelines, there are several common themes that emerge across various

60
00:04:48,010 --> 00:04:48,910
frameworks.

61
00:04:49,390 --> 00:04:56,290
These include the importance of human oversight, transparency, fairness, privacy, and accountability.

62
00:04:56,470 --> 00:05:02,970
For instance, the principle of transparency is emphasized in both the EU and OECD guidelines, highlighting

63
00:05:02,970 --> 00:05:07,110
the need for AI systems to be understandable and explainable to users.

64
00:05:07,530 --> 00:05:13,980
Similarly, the principle of fairness is a common thread, with guidelines from the EU, China, Japan

65
00:05:13,980 --> 00:05:19,410
and the OECD all stressing the need to prevent biases and discrimination in AI systems.

66
00:05:19,890 --> 00:05:25,020
These commonalities reflect a shared understanding of the fundamental ethical issues associated with

67
00:05:25,020 --> 00:05:27,960
AI, and the need to address them comprehensively.

68
00:05:29,220 --> 00:05:34,710
However, there are also notable differences that reflect the unique cultural, social, and political

69
00:05:34,710 --> 00:05:36,660
contexts of different regions.

70
00:05:37,260 --> 00:05:42,960
The EU's guidelines, for instance, place a strong emphasis on individual rights and data protection,

71
00:05:42,960 --> 00:05:47,220
reflecting the region's robust legal framework for privacy and human rights.

72
00:05:47,580 --> 00:05:53,760
In contrast, China's guidelines prioritize societal stability and national security, highlighting

73
00:05:53,760 --> 00:05:56,430
the government's role in guiding AI development.

74
00:05:57,360 --> 00:06:02,710
The EU's approach, with its emphasis on innovation and market driven solutions, Allusions contrasts

75
00:06:02,710 --> 00:06:05,890
with the more regulatory approaches of the EU and China.

76
00:06:06,760 --> 00:06:12,340
Japan's human centric approach, emphasizing harmony and coexistence, reflects the country's cultural

77
00:06:12,340 --> 00:06:14,410
values and societal priorities.

78
00:06:15,550 --> 00:06:20,050
These differences have significant implications for the global governance of AI.

79
00:06:20,440 --> 00:06:25,690
The varying approaches to AI ethics can lead to challenges in achieving international alignment and

80
00:06:25,690 --> 00:06:26,710
cooperation.

81
00:06:27,190 --> 00:06:33,220
For instance, the emphasis on data privacy in the EU may conflict with the more lenient data practices

82
00:06:33,220 --> 00:06:37,150
in other regions, leading to potential regulatory clashes.

83
00:06:37,780 --> 00:06:44,050
Similarly, the different priorities in AI development, such as the US's focus on innovation versus

84
00:06:44,050 --> 00:06:49,240
China's focus on societal stability, can lead to divergent paths in AI governance.

85
00:06:49,870 --> 00:06:55,630
These challenges underscore the need for ongoing dialogue and collaboration to harmonize ethical standards

86
00:06:55,630 --> 00:07:00,460
and ensure the responsible development and deployment of AI technologies globally.

87
00:07:01,990 --> 00:07:07,760
To illustrate the practical implications of these ethical guidelines, consider the case of facial recognition

88
00:07:07,760 --> 00:07:09,980
technology in the EU.

89
00:07:10,010 --> 00:07:15,920
The use of facial recognition technology is heavily regulated to protect individual privacy and prevent

90
00:07:15,920 --> 00:07:16,940
discrimination.

91
00:07:17,390 --> 00:07:23,810
The General Data Protection Regulation imposes strict requirements on the use of biometric data, ensuring

92
00:07:23,810 --> 00:07:26,120
that individual's rights are protected.

93
00:07:26,570 --> 00:07:33,410
In contrast, the use of facial recognition technology in China is widespread and integrated into various

94
00:07:33,440 --> 00:07:37,850
aspects of public life, from surveillance to social credit systems.

95
00:07:38,090 --> 00:07:42,170
This reflects the country's emphasis on societal stability and security.

96
00:07:42,710 --> 00:07:49,250
In the US, the use of facial recognition technology varies by sector and jurisdiction, with some states

97
00:07:49,250 --> 00:07:53,540
imposing strict regulations while others adopt a more permissive approach.

98
00:07:53,570 --> 00:08:00,020
These differences highlight the diverse ethical considerations and regulatory approaches to AI technologies

99
00:08:00,020 --> 00:08:01,670
across different regions.

100
00:08:02,920 --> 00:08:10,360
In conclusion, the comparison of global ethical guidelines for AI reveals both commonalities and differences

101
00:08:10,360 --> 00:08:15,790
that reflect the unique cultural, social, and political contexts of different regions.

102
00:08:16,360 --> 00:08:21,850
While there is a shared understanding of the fundamental ethical issues associated with AI, the varying

103
00:08:21,850 --> 00:08:28,360
approaches to addressing these issues highlight the challenges in achieving global alignment in AI ethics.

104
00:08:28,390 --> 00:08:34,330
These differences underscore the importance of ongoing dialogue and collaboration to harmonize ethical

105
00:08:34,330 --> 00:08:39,880
standards and ensure the responsible development and deployment of AI technologies globally.

106
00:08:40,510 --> 00:08:46,990
As AI continues to evolve and permeate various aspects of society, the need for robust and coherent

107
00:08:46,990 --> 00:08:50,290
ethical guidelines becomes increasingly critical.

108
00:08:50,920 --> 00:08:56,110
By examining and understanding these global ethical guidelines, we can better navigate the complex

109
00:08:56,110 --> 00:09:02,620
landscape of AI ethics and contribute to the development of trustworthy and responsible AI systems.
