1
00:00:00,050 --> 00:00:00,410
Lesson.

2
00:00:00,410 --> 00:00:01,970
Human centric AI systems.

3
00:00:01,970 --> 00:00:08,540
Human centric AI systems prioritize human values, ethics, and well-being in the development and deployment

4
00:00:08,540 --> 00:00:10,130
of artificial intelligence.

5
00:00:10,820 --> 00:00:16,640
Understanding the importance of human centric AI is crucial for AI governance professionals tasked with

6
00:00:16,640 --> 00:00:21,260
ensuring that AI technologies are developed responsibly and are trustworthy.

7
00:00:22,010 --> 00:00:28,220
This lesson aims to provide a comprehensive understanding of human centric AI systems within the context

8
00:00:28,250 --> 00:00:31,550
of responsible AI principles and trustworthy AI.

9
00:00:31,580 --> 00:00:37,940
Integrating relevant statistics, examples, and scholarly insights to substantiate key points.

10
00:00:37,940 --> 00:00:44,480
Human centric AI emphasizes the need for AI systems to align with human values and ethics, ensuring

11
00:00:44,480 --> 00:00:50,360
that these systems not only perform tasks efficiently, but also respect human dignity, rights, and

12
00:00:50,360 --> 00:00:51,590
societal norms.

13
00:00:52,130 --> 00:00:58,640
This approach contrasts starkly with purely utilitarian or efficiency driven AI models, which may overlook

14
00:00:58,640 --> 00:01:01,310
the broader impact on individuals and communities.

15
00:01:01,340 --> 00:01:07,200
Research has shown that AI systems designed without considering human values can lead to unintended

16
00:01:07,200 --> 00:01:13,740
consequences such as privacy infringements, biased decision making, and erosion of trust in technology.

17
00:01:15,090 --> 00:01:20,490
One of the fundamental aspects of human centric AI is the integration of ethical principles into the

18
00:01:20,490 --> 00:01:23,430
AI life cycle, from design to deployment.

19
00:01:23,760 --> 00:01:29,700
Ethical AI frameworks often include principles such as fairness, accountability, transparency, and

20
00:01:29,700 --> 00:01:30,600
privacy.

21
00:01:31,020 --> 00:01:37,170
For instance, the EU's Ethics Guidelines for trustworthy AI outline seven key requirements human agency

22
00:01:37,170 --> 00:01:43,140
and oversight, technical robustness and safety, privacy and data governance, transparency, diversity,

23
00:01:43,170 --> 00:01:45,180
non-discrimination and fairness.

24
00:01:45,180 --> 00:01:48,930
Societal and environmental well-being and accountability.

25
00:01:49,500 --> 00:01:55,110
These guidelines serve as a blueprint for developing AI systems that respect human values and promote

26
00:01:55,110 --> 00:01:56,190
societal good.

27
00:01:57,210 --> 00:02:03,450
Fairness in AI involves ensuring that AI systems do not perpetuate or exacerbate existing biases, and

28
00:02:03,450 --> 00:02:04,530
inequalities.

29
00:02:04,890 --> 00:02:11,410
Studies have shown that AI algorithms can inherit biases present in training data, leading to discriminatory

30
00:02:11,410 --> 00:02:12,310
outcomes.

31
00:02:12,670 --> 00:02:18,910
For example, a widely cited case is the Compas algorithm used in the US criminal justice system, which

32
00:02:18,910 --> 00:02:24,340
was found to be biased against African American defendants, leading to higher false positive rates

33
00:02:24,340 --> 00:02:25,990
compared to white defendants.

34
00:02:26,560 --> 00:02:32,350
Addressing fairness requires a multifaceted approach, including diverse data representation, bias

35
00:02:32,350 --> 00:02:37,720
detection and mitigation techniques, and continuous monitoring and evaluation of AI systems.

36
00:02:39,160 --> 00:02:43,300
Transparency is another critical component of human centric AI.

37
00:02:43,690 --> 00:02:49,780
Transparency involves providing clear and understandable explanations of how AI systems make decisions,

38
00:02:49,780 --> 00:02:54,460
thereby enabling users to trust and effectively interact with these systems.

39
00:02:55,090 --> 00:03:00,700
The concept of explainable AI has gained prominence in recent years, aiming to make AI decision making

40
00:03:00,700 --> 00:03:04,510
processes more interpretable and understandable to humans.

41
00:03:05,170 --> 00:03:10,690
For instance, in the healthcare domain, explainable AI can help clinicians understand the reasoning

42
00:03:10,690 --> 00:03:14,410
behind AI generated diagnoses or treatment Recommendations.

43
00:03:14,410 --> 00:03:17,860
Fostering trust and facilitating better decision making.

44
00:03:19,150 --> 00:03:25,630
Accountability in AI involves establishing mechanisms to hold AI developers and deployers responsible

45
00:03:25,630 --> 00:03:27,490
for the outcomes of their systems.

46
00:03:28,240 --> 00:03:34,090
This includes defining clear roles and responsibilities, implementing robust governance frameworks,

47
00:03:34,090 --> 00:03:37,660
and ensuring compliance with legal and ethical standards.

48
00:03:38,470 --> 00:03:43,900
The General Data Protection Regulation in the European Union, for example, includes provisions for

49
00:03:43,900 --> 00:03:49,540
algorithmic accountability, granting individuals the right to challenge and seek explanations for automated

50
00:03:49,540 --> 00:03:52,030
decisions that significantly affect them.

51
00:03:52,750 --> 00:03:58,390
Such regulatory measures are essential for safeguarding individuals rights and promoting responsible

52
00:03:58,420 --> 00:03:59,410
AI use.

53
00:04:01,120 --> 00:04:06,550
Privacy is a fundamental human right that must be protected in the development and deployment of AI

54
00:04:06,580 --> 00:04:07,360
systems.

55
00:04:07,900 --> 00:04:13,270
AI technologies often rely on vast amounts of personal data, raising concerns about data security and

56
00:04:13,270 --> 00:04:14,590
privacy breaches.

57
00:04:15,040 --> 00:04:21,180
Ensuring robust data protection measures such as encryption, anonymization and secure data storage

58
00:04:21,180 --> 00:04:25,470
is crucial for maintaining user trust and complying with legal requirements.

59
00:04:26,070 --> 00:04:32,220
Moreover, privacy preserving AI techniques such as federated learning and differential privacy can

60
00:04:32,220 --> 00:04:36,930
enable the development of AI models without compromising individual privacy.

61
00:04:38,340 --> 00:04:43,860
Human centric AI also emphasizes the importance of societal and environmental well-being.

62
00:04:44,490 --> 00:04:49,800
AI systems should be designed to contribute positively to society and the environment, addressing global

63
00:04:49,800 --> 00:04:53,370
challenges such as climate change, health care, and education.

64
00:04:53,970 --> 00:05:00,360
For example, I can be leveraged to optimize energy consumption, reduce carbon emissions, and improve

65
00:05:00,360 --> 00:05:04,200
resource management, contributing to environmental sustainability.

66
00:05:04,890 --> 00:05:11,160
In healthcare, AI powered diagnostic tools and personalized treatment plans can enhance patient outcomes

67
00:05:11,160 --> 00:05:15,420
and access to medical services, particularly in underserved regions.

68
00:05:16,320 --> 00:05:19,290
Trust is a cornerstone of human centric AI.

69
00:05:19,650 --> 00:05:24,950
Building and maintaining trust in AI systems requires addressing the aforementioned principles fairness,

70
00:05:24,980 --> 00:05:31,580
transparency, accountability and privacy, while also fostering public engagement and education.

71
00:05:32,000 --> 00:05:37,190
Public understanding and acceptance of AI technologies are crucial for their successful integration

72
00:05:37,190 --> 00:05:38,270
into society.

73
00:05:38,450 --> 00:05:44,510
Engaging with diverse stakeholders, including policymakers, industry leaders, academics, and civil

74
00:05:44,510 --> 00:05:49,850
society can help ensure that AI systems reflect a broad range of perspectives and values.

75
00:05:51,260 --> 00:05:57,410
In conclusion, human centric AI systems are essential for aligning AI development with human values,

76
00:05:57,410 --> 00:06:04,400
ethics, and societal well-being by incorporating principles such as fairness, transparency, accountability,

77
00:06:04,400 --> 00:06:10,190
and privacy into the AI life cycle, we can develop AI technologies that are not only effective, but

78
00:06:10,190 --> 00:06:12,830
also trustworthy and beneficial to society.

79
00:06:13,610 --> 00:06:18,650
As AI governance professionals, it is our responsibility to advocate for and implement human centric

80
00:06:18,650 --> 00:06:24,920
AI practices while ensuring that AI systems are developed and deployed in a manner that respects human

81
00:06:24,920 --> 00:06:27,380
dignity and promotes the common good.