AI Ethics.pdf

SWARNPRASTHA PUBLIC SCHOOL ARTIFICIAL INTELLIGENCE/2024-25 UNIT 1- AI REFLECTION, AI PROJECT LIFE CYCLE & AI ETHICS SESSION 3 – ETHICS & MORALITY Ethics: are the moral responsibility of anyo...

SWARNPRASTHA PUBLIC SCHOOL ARTIFICIAL INTELLIGENCE/2024-25 UNIT 1- AI REFLECTION, AI PROJECT LIFE CYCLE & AI ETHICS SESSION 3 – ETHICS & MORALITY Ethics: are the moral responsibility of anyone or anything that can impact others. AI Ethics: It is a set of values, principles and techniques that employ widely accepted standards of right & wrong to guide moral conduct in the development and use of AI Technologies. Major Ethical Issues of AI - Safety AI Actions: -Human AI Interaction -Cyber Security & malicious use -Trust Privacy & Control AI Setup -Bias & Fairness -Accountability - Transparency AI Impact - Automation - Imapact over jobs - Auditibilty -Interpretability AI Future - Control of AI -Human Rightd V/S Robot Rights.  Bias & Fairness: AI system should be free from all types of biases and be fair.  Accountability: AI learns and evolves around time and data. What if an evolved algorithm commits some mistakes? Who would be accountable for it?  Transparency: It means nothing should be hidden and whatever actions are performed by AI should be explainable.  Safety: AI should be implemented in such a way that it will not cause direct or indirect harm to data, people and the outcomes.  Human AI Interaction: AI model must not deceive humans or other living beings, and it must not threaten or violate human dignity in any way.  Trust, Privacy & Control: As the AI models are improving, the technology make what one reliable evidence into unreliable evidence ex- digital photos, sound recordings and video. With the “Deepfake” it’s very easy to create rather than alter the photos. Thus, it is the ethical responsibility of the creators and user to AI to ensure that these are not misused. Deepfake is a technology that can generate fake digital photos, sound recordings and video, which look just as original as possible.  Cyber Security & Malicious use: Since AI uses big datasets to determine futuristic threats and to test the system’s vulnerabilities. If such tools and datasets become available to hackers then it will be for cyber threats. Thus, it is the ethical responsibility of an organization to have a human control over AI usage so that it not available to hackers.  Automation & Impact Over Jobs: AI & Robotics are increasing in automation in all types of fields and industries, leading job loss for human being. But AI does not mean that jobs are reduced, it just means that the nature of job and work is changing. So it is the ethical responsibility of the government & other organizations to upgrade the skills of their employees.  Human Rights in the age of AI: AI has generated new forms of threats like how to protect human rights in the age of AI? Example- In smart cities, with the help of various applications all the data of the user can be accessed leading to a huge risk of data privacy and protection, violating human right to privacy. AI BIAS AND AI ACCESS  AI Bias: Bias means inclination or prejudice for or against one person or group, especially in a way considered to be unfair. When AI programs, tools and algorithms exhibit any kind of bias is called AI Bias. Ex- A health insurance AI program preferred whites over blacks for extra healthcare services because of an AI Bias. Reasons for AI Bias in Data  Human Bias in decisions  Flawed & unbalanced data collection  Under or over representation of specific features.  Wrong assumptions  No proper bias testing  How to ensure data fairness?  Identify the correlation of features with data.  Ensuring balanced data  Observing biases in human decisions and the collected data.  Supervised Decision Making Some Trusted AI Principles  Responsible: Safeguarding human rights and protecting the data we are entrusted with.  Accountable: Seeking and levering feedback for continuous improvement.  Transparent: Developing a transparent user experience to guide users through machine driven recommendations.  Empowering: Promoting economic growth and employment for our customers, their employees, and society as a whole.  Inclusive: Respecting the societal value of all those impacted, not just those of the creators. How to Reduce and Mitigating AI Bias  Through Research  Diversity of Team  Data Diversity  Standardized Data Labelling  Identifying Bias proneness  Data Review  Regular Data Analysis  Regular Bias Testing x------------------------------------x----------------------------------------x