1
00:00:00,080 --> 00:00:02,930
OECD and EU standards for trustworthy AI.

2
00:00:02,930 --> 00:00:08,720
The organisation for Economic Co-operation and Development and the European Union have both established

3
00:00:08,720 --> 00:00:12,290
comprehensive frameworks for trustworthy artificial intelligence.

4
00:00:12,590 --> 00:00:18,680
These frameworks are designed to foster innovation while ensuring that AI systems are ethical, transparent

5
00:00:18,680 --> 00:00:20,600
and aligned with human values.

6
00:00:21,050 --> 00:00:28,670
The OECD's AI principles adopted in 2019 emphasize inclusiveness, transparency, accountability, robustness,

7
00:00:28,670 --> 00:00:30,290
security, and safety.

8
00:00:30,830 --> 00:00:36,560
Similarly, the EU's Ethics Guidelines for trustworthy AI published by the High Level Expert Group on

9
00:00:36,560 --> 00:00:43,400
AI in 2019, stress respect for human autonomy, prevention of harm, fairness and explicability.

10
00:00:44,060 --> 00:00:49,010
Together, these standards form a robust foundation for responsible AI governance.

11
00:00:50,000 --> 00:00:53,870
The OECD AI principles are anchored in five key areas.

12
00:00:54,230 --> 00:00:59,830
First, I should benefit people and the planet by driving inclusive growth, sustainable development

13
00:00:59,830 --> 00:01:00,820
and wellbeing.

14
00:01:01,210 --> 00:01:05,980
This principle underscores the importance of ensuring that AI technologies contribute positively to

15
00:01:06,010 --> 00:01:09,610
society and do not exacerbate existing inequalities.

16
00:01:10,180 --> 00:01:15,940
Second, AI systems should be designed in a way that respects the rule of law, human rights, democratic

17
00:01:15,940 --> 00:01:17,590
values, and diversity.

18
00:01:18,010 --> 00:01:24,520
This includes principles of fairness and non-discrimination, ensuring that AI does not perpetuate biases

19
00:01:24,520 --> 00:01:26,740
or infringe upon individual rights.

20
00:01:27,460 --> 00:01:32,020
Third, there is a strong emphasis on transparency and responsible disclosure.

21
00:01:32,410 --> 00:01:38,200
This principle argues that AI systems should be transparent and explainable, allowing users to understand

22
00:01:38,200 --> 00:01:40,870
and trust the decisions made by these systems.

23
00:01:41,320 --> 00:01:47,440
Fourth, AI systems must function in a robust, secure and safe manner throughout their life cycles.

24
00:01:47,470 --> 00:01:53,160
This includes not only technical robustness and safety, but also the capacity to deal with errors or

25
00:01:53,160 --> 00:01:54,540
inconsistencies.

26
00:01:55,200 --> 00:02:01,200
Finally, organizations and individuals developing, deploying, or operating AI systems should be held

27
00:02:01,200 --> 00:02:05,070
accountable for their proper functioning in line with the above principles.

28
00:02:06,870 --> 00:02:13,530
The EU's ethics guidelines for trustworthy AI build on similar foundations but offer additional granularity.

29
00:02:13,530 --> 00:02:17,070
They propose a framework based on seven key requirements.

30
00:02:17,400 --> 00:02:23,790
These include human agency and oversight, which ensures that AI systems support human autonomy and

31
00:02:23,790 --> 00:02:24,840
decision making.

32
00:02:25,320 --> 00:02:30,900
Technical robustness and safety are also highlighted, emphasizing resilience to attack and security

33
00:02:30,930 --> 00:02:35,250
fallbacks in case of failure, and accurate and reliable AI systems.

34
00:02:35,250 --> 00:02:41,130
Privacy and data governance are crucial, ensuring that personal data is protected and used responsibly.

35
00:02:41,460 --> 00:02:47,970
Transparency, including traceability, explainability and communication, is another key requirement

36
00:02:47,970 --> 00:02:52,060
ensuring that AI operations are comprehensible to all stakeholders.

37
00:02:52,300 --> 00:02:58,330
Diversity, non-discrimination and fairness guarantee that AI systems are accessible and do not harm

38
00:02:58,330 --> 00:03:00,430
or discriminate against any groups.

39
00:03:00,430 --> 00:03:06,880
Societal and environmental well-being focuses on ensuring that AI systems contribute positively to society

40
00:03:06,880 --> 00:03:07,960
and the environment.

41
00:03:08,380 --> 00:03:14,770
Finally, accountability mechanisms should be in place ensuring that organizations implementing AI are

42
00:03:14,770 --> 00:03:16,210
answerable for their systems.

43
00:03:16,210 --> 00:03:17,110
Outcomes.

44
00:03:18,280 --> 00:03:24,160
The integration of these principles and guidelines into AI governance frameworks is not merely theoretical,

45
00:03:24,160 --> 00:03:26,170
but has practical implications.

46
00:03:26,440 --> 00:03:32,320
For instance, the GDPR in the EU has provisions that overlap with AI ethics, particularly concerning

47
00:03:32,320 --> 00:03:34,120
data privacy and protection.

48
00:03:34,690 --> 00:03:40,870
GDPR mandates that individuals have the right to explanation when subjected to automated decision making,

49
00:03:40,870 --> 00:03:46,800
reflecting the transparency and accountability principles outlined in the EU guidelines.

50
00:03:47,220 --> 00:03:52,920
Moreover, the OECD and EU frameworks have influenced regulatory developments in other regions.

51
00:03:53,400 --> 00:03:58,800
For example, the Canadian Directive on Automated Decision Making incorporates similar principles,

52
00:03:58,800 --> 00:04:03,630
ensuring that government use of AI is transparent, equitable and accountable.

53
00:04:05,160 --> 00:04:08,850
Statistics illustrate the growing importance of these frameworks.

54
00:04:09,600 --> 00:04:15,960
As of 2020, over 60 countries have adopted national AI strategies, many of which align closely with

55
00:04:15,960 --> 00:04:18,150
the OECD and EU guidelines.

56
00:04:18,660 --> 00:04:24,360
This global adoption underscores the relevance and applicability of these standards in guiding responsible

57
00:04:24,390 --> 00:04:26,280
AI development and deployment.

58
00:04:27,060 --> 00:04:33,930
Furthermore, a 2021 survey found that 76% of organizations consider ethics to be a crucial factor in

59
00:04:33,930 --> 00:04:39,600
their AI strategies, reflecting the increasing recognition of the importance of trustworthy AI.

60
00:04:41,480 --> 00:04:47,120
Examples of the application of these principles can be seen in various sectors in healthcare.

61
00:04:47,150 --> 00:04:53,180
AI systems are being developed to assist in diagnostic processes, treatment recommendations, and patient

62
00:04:53,180 --> 00:04:53,930
management.

63
00:04:54,380 --> 00:05:00,320
These systems must adhere to the principles of transparency, accountability, and fairness to ensure

64
00:05:00,320 --> 00:05:03,380
that they provide accurate and unbiased information.

65
00:05:03,890 --> 00:05:09,920
For instance, IBM's Watson for oncology has been designed to support oncologists by providing evidence

66
00:05:09,950 --> 00:05:11,390
based treatment options.

67
00:05:11,420 --> 00:05:17,420
However, it has faced scrutiny regarding its transparency and accuracy, highlighting the need for

68
00:05:17,420 --> 00:05:19,400
robust governance frameworks.

69
00:05:19,850 --> 00:05:25,100
In finance, AI is used for credit scoring, fraud detection and algorithmic trading.

70
00:05:25,430 --> 00:05:31,490
Ensuring that these systems are fair, transparent and accountable is crucial to maintaining trust and

71
00:05:31,490 --> 00:05:32,900
preventing discrimination.

72
00:05:32,930 --> 00:05:38,800
The EU's ethics guidelines have been influential in shaping regulations that govern the use of AI in

73
00:05:38,800 --> 00:05:44,140
financial services, ensuring that these systems respect human rights and democratic values.

74
00:05:44,680 --> 00:05:45,370
Baths.

75
00:05:45,790 --> 00:05:51,940
The development and implementation of trustworthy AI are ongoing processes that require continuous monitoring,

76
00:05:51,970 --> 00:05:54,010
evaluation, and adaptation.

77
00:05:54,430 --> 00:06:00,670
The OECD and EU frameworks provide a solid foundation, but the dynamic nature of AI technology necessitates

78
00:06:00,670 --> 00:06:03,010
that these guidelines evolve over time.

79
00:06:03,460 --> 00:06:09,220
Stakeholder engagement, including input from industry, academia, civil society, and the public,

80
00:06:09,220 --> 00:06:13,390
is essential to ensure that these frameworks remain relevant and effective.

81
00:06:14,230 --> 00:06:20,230
Additionally, international cooperation is crucial to harmonize standards and promote global adherence

82
00:06:20,230 --> 00:06:22,030
to ethical AI principles.

83
00:06:22,480 --> 00:06:28,600
This collaboration is particularly important as AI systems often operate across borders, and inconsistencies

84
00:06:28,600 --> 00:06:37,040
in regulations can lead to ethical and legal challenges In conclusion, the OECD and EU standards for

85
00:06:37,040 --> 00:06:42,830
trustworthy AI provide a comprehensive and robust framework for ensuring that AI systems are developed

86
00:06:42,830 --> 00:06:44,510
and deployed responsibly.

87
00:06:44,870 --> 00:06:51,170
These standards emphasize principles such as transparency, accountability, fairness and safety, which

88
00:06:51,170 --> 00:06:54,650
are crucial for maintaining public trust and preventing harm.

89
00:06:55,430 --> 00:07:00,530
The practical implications of these frameworks are evident in various sectors, including healthcare

90
00:07:00,530 --> 00:07:06,680
and finance, where AI systems must adhere to ethical guidelines to ensure their positive impact.

91
00:07:07,370 --> 00:07:12,200
The global adoption of these principles, and the increasing recognition of the importance of trustworthy

92
00:07:12,230 --> 00:07:16,250
AI underscore the relevance and applicability of these standards.

93
00:07:17,240 --> 00:07:19,550
As AI technology continues to evolve.

94
00:07:19,580 --> 00:07:25,130
Ongoing monitoring, evaluation, and international cooperation will be essential to ensure that these

95
00:07:25,130 --> 00:07:27,470
frameworks remain effective and relevant.
