Introduction to Ethics: Theories and Principles

Introduction to Ethics

• Ethics is the study of morality
• Morality refers to the system of rules that guides human conduct and includes principles for evaluating these rules
• A moral system aims to prevent harm and promote the flourishing of human beings
• Moral systems should be public, informal, rational, and impartial
• Morality is functional at both the micro-level (individual behavior) and macro-level (social policy)

Theories of Ethics: Divine Command Ethics

• The morality of an action is determined by divine rules
• Morality is dictated by a superior power such as God or karma
• A criticism of divine command ethics is the question of what happens if no divine law-giver exists

Theories of Ethics: Consequentialist Ethics / Utilitarianism

• Morality hinges on the outcome of an action
• Methods to evaluate an action can include the happiness of those affected
• Methods to evaluate an action can include health and well-being of those affected
• A criticism is whether morality can be reduced to happiness, health, or well-being

Theories of Ethics: Deontological Ethics

• Morality hinges on duty instead of outcomes
• Immanuel Kant's Moral Imperative states to act on maxims that can be universally applied
• A criticism is what happens when duties conflict

Theories of Ethics: Rights and Social Contract Ethics

• Morality is defined through upholding human rights and following social contracts
• A criticism is whether all moral issues can be covered by contracts or rights

Theories of Ethics: Virtue Ethics

• Focuses on character and virtue rather than rules or consequences
• It can align with other ethical theories
• Virtues differ by culture and context

Moral Relativism vs. Objectivity

• Moral relativism says that morality is subjective and varies across cultures
• Cultural relativism observes differences but does not dictate moral judgement
• Ethical objectivity says that ethics can be rational even if morality isn't absolute

Moral Foundations Theory (Jonathan Haidt)

• Human morality has six core domains
• Care/Harm (Compassion): feeling empathy for others and wanting to prevent suffering; example: helping a child who is lost or donating to disaster relief efforts; typically emphasized in liberal moral views, human rights advocacy, and healthcare ethics
• Fairness/Cheating (Justice): valuing justice, equality, and fair treatment; example: supporting equal pay for equal work or being outraged by corruption and cheating; often emphasized in social justice, economic fairness, and discussions of democracy
• Liberty/Oppression (Freedom): opposing domination and valuing personal and collective freedom; example: fighting against dictatorships, supporting free speech, or advocating for civil liberties
• Loyalty/Betrayal (Group Identity): feeling strong connections to groups and valuing loyalty; example: patriotism, supporting sports teams, or being upset by a friend's betrayal; often emphasized in conservative political ideologies, military culture, and collectivist socieites
• Authority/Subversion (Respect for Hierarchy): recognizing and respecting hierarchies and traditions; example: following the chain of command in the military or respecting elders; typically found in conservative politics, traditional societies and religious institutions
• Sanctity/Degradation (Purity and Sacredness): seeing some things as sacred and rejecting what is viewed as disgusting or degrading; example: religious rituals, disgust at immoral behavior, or opposition to taboo acts like incest or cannibalism; common in religious and traditionalist perspectives and movements focused on moral purity
• Societies emphasize different foundations such as liberal vs. conservative ethics

Just-Consequentialist Framework (Moor)

• Combines multiple ethical theories, except divine command
• Steps for policy evaluation: 1. deliberate impartially over options; 2. a policy is ethical if it avoids unnecessary harm and supports fulfilling duties and obligations; 3. identify the best policy by weighing benefits and harms and ensuring factual accuracy; 4. consult stakeholders for ethical decision-making

Ethical Issues in AI and Autonomous Vehicles

• Autonomous cars can prioritize passenger safety vs. pedestrian safety
• Autonomous cars can prioritize protecting property vs. preventing harm
• A utilitarian approach is to minimize harm
• A deontological approach is to follow universal moral rules
• Virtue ethics is to act with wisdom and justice

Introduction to Fairness in Ethics

• Fairness is a key principle in ethical discussions
• Fairness ensures just treatment for all individuals
• The GAIA Policy emphasizes fairness in education and AI use
• Fairness focuses on transparency, equity, and responsible AI usage
• Key questions to consider include what is fairness, how to evaluate fairness, and whether to account for personal differences

The Importance of Fairness

• Fairness supports social stability (John Rawls)
• Perceived unfairness can lead to societal unrest or revolution
• Fairness upholds individual dignity by preventing discrimination
• Cultural differences impact the perception of fairness
• In some societies, pay gaps based on gender are considered fair due to cultural norms

Equality vs. Equity

• Equality is treating everyone the same, regardless of differences
• Equity is adjusting conditions so everyone has an equal opportunity
• Equity involves giving a short person a taller box to see over a fence, while equality involves both people getting the same sized box to stand on

Fairness in Social Policy (Distributive Justice)

• Distributive justice seeks to ensure the fair distribution of goods and opportunities
• Distributive justice is based on what is being distributed such as jobs, healthcare, and education
• Distributive justice is based on who receives it such as individuals, groups, and regions
• Distributive justice is based on equality, merit, and past discrimination
• Distributive justice raises the question of government providing universal healthcare
• Distributive justice raises the question of CEOs receiving extremely high salaries
• Distributive justice raises the question of women receiving equal pay for equal work

Distribution of Goods and Harms

• Types of Goods: Environmental (clean air, water); Economic (jobs, income, property); Personal (education, healthcare, privacy); Social (peace, safety, justice)
• Types of Harms: Distributional harms (unequal distribution of goods); Representational harms (stereotyping or misrepresentation)
• Privilege affects fairness
• Privileged individuals access goods more easily and avoid harms more effectively

Algorithmic Fairness

• Bias in AI systems affects fairness
• Amazon's AI recruiting tool discriminated against women because it was trained on biased data
• Unintended bias can exist in algorithms even if trained on “neutral” data
• AI trained to detect sexual orientation from photos showed biased accuracy across groups

The Limits of Blinding Data to Improve Fairness

• "Fairness through unawareness" does not work
• AI can infer race, gender, or socioeconomic status from data such as ZIP codes or shopping habits
• Sometimes, fairness requires using sensitive data
• Medical treatments may need race/gender data for accurate prescriptions
• Recidivism risk algorithms may need gender information, as men and women have different reoffending patterns

Group Fairness & Systemic Bias

• Fairness definitions: Demographic Parity (everyone gets equal treatment); Error-Rate Parity (different groups face equal misclassification rates); Calibration (algorithm accuracy is equal across groups)
• Systemic Bias Examples: AP classes are easier to access in wealthy schools, so using "number of AP classes" as a success metric is unfair; Sound compression in technology often favors men's voices over women's

Fairness and the Digital Divide

• Unequal access to information and technology creates fairness concerns
• Developing nations have less internet access
• Economic class, race, and gender affect access to jobs, education, and information

The Role of Cyber Technology in Fairness

• Technology enhances free speech and informs voters
• Algorithmic personalization creates echo chambers with biased news feeds
• Technology can undermine democracy by spreading misinformation
• Job displacement occurs due to AI and automation
• Increased workplace surveillance impacts privacy, job security, and well-being

Key Takeaways on Fairness

• Fairness is complex and culturally dependent
• Equity, not just equality, is needed for real fairness
• Algorithmic fairness is difficult but crucial to achieve
• Bias exists in both human and AI decision-making
• Technology influences fairness in democracy, work, and access to resources

Introduction to Ethics and Social Goods

• A responsible computing and data scientist should use data and algorithms for social good
• A responsible computing and data scientist should protect privacy, dignity, and agency of users
• A responsible computing and data scientist should recognize the long-term impact of technology
• A responsible computing and data scientist should seek expert help when necessary
• Key questions include what is public good vs. public harm, how to evaluate social good and public good, and who decides what qualifies as public good or harm

Understanding Goods and Harms

• Types of Goods: Environmental (clean air, water, food safety); Economic (jobs, property, mobility); Personal (healthcare, education, privacy); Social (peace, stability, safety, justice)
• Types of Harms: Distributional harms (inequitable access to goods); Representational harms (stereotyping, misrepresentation)
• Those with privilege access goods more efficiently and avoid harms more effectively

Definitions of Key Terms

• Public Goods: a commodity/service that benefits all members of a society
• Public Goods have non-excludable qualities (everyone can access it)
• Public Goods have non-rivalrous qualities (one person's use doesn't reduce another's)
• Pure public goods are always non-excludable such as clean air or national defense
• Impure public goods are sometimes non-excludable such as open-source software
• Private Good: Excludable and rivalrous (e.g., food, personal computers)
• Club Good: Excludable but non-rivalrous (e.g., satellite TV, toll roads)
• Common-Pool Good: Non-excludable but rivalrous (e.g., fisheries, natural resources)

Social Goods vs. Public Goods

• Social Goods are a subset of public goods related to societal well-being such as education and public safety
• Public Health is a complex system that affects everyone
• Healthcare should be a universal right, according to some political movements
• Public-Private Partnerships (PPPs)
• Governments provide funding for PPPs
• Private companies provide expertise for PPPs
• An example of a PPP is the development of the internet and COVID vaccines

The Ethics of Generating Public Goods

• Public Good Game (Illustration): everyone starts with $100 and people contribute to a shared public good ($3,000 needed); If the goal is met, each person gets $200, but if not met, they lose the money
• Outcomes: some free-ride and gain benefits without contributing, some cooperate fully and ensure public good is created, some cooperate equitably and contribute based on ability, some appoint leaders to manage contributions
• Free-riding prevents public goods from being created
• Solutions to Free-Riding: punishment for free riders (but costly to enforce), cultural indoctrination (teaching cooperation), religious/moral beliefs (invisible moral judge)
• Challenges in Public Goods Production: Free markets often fail to provide public goods; Governments may fail due to political disagreement; Economic systems need virtuous behavior but cannot create it

Cyber-Technology and Public Goods

• Cyber-tech is difficult to create as a public good due to high cost and profit incentives; companies often have no intrinsic reason to make tech non-excludable
• Public-Private Partnerships for Cyber-Tech: Governments have the motivation but lack expertise; Corporations have the expertise but prioritize profit

The Challenge of Transforming Cyber-Tech into Public Goods

• Are cyber technologies more often public goods or public harms?
• How can we ensure corporations, governments, and NGOs transform technology into public goods?
• Should cyber-tech public goods be regulated and if so, by whom?
• Examples of Cyber-Tech Public Goods: Internet & search engines; Social networking services; Virtual environments; AI algorithms

Key Takeaways on Social Goods and Ethics

• Public goods benefit all but are hard to create due to free-riding and profit motives
• Social goods focus on societal organization and function, such as public health and safety
• Markets often fail to produce public goods, requiring government or collective action
• Cyber-tech could be a public good, but private corporations resist making it non-excludable
• Public-private partnerships may be the best approach to creating cyber-tech public goods

Introduction to Ethical Reasoning and Cognitive Errors

• Ethical judgments are made by human minds, which are shaped by evolution and cognitive psychology
• Evolutionary Biology: Our brains evolved for survival, not perfect rationality
• Cognitive Psychology: Reveals how we process thoughts, emotions, and decisions
• Cognitive Limitations: Many cognitive features that were once adaptive can now be maladaptive in modern contexts

Cognitive Bias: Definition and Importance

• Bounded Rationality: Humans have limited cognitive resources to process complex problems
• Cognitive Bias: Systematic thinking errors that affect judgments and decisions
• Origins of Bias: Heuristics (mental shortcuts) to deal with limited data, neuro-evolutionary constraints (our brains prioritize survival-based reasoning)
• Cognitive biases distort logical reasoning, probability evaluation, and judgment
• Biases arise because our brains were designed for fast, survival-based decisions rather than complex logical analysis

Types of Cognitive Biases

• Cognitive biases occur due to four main reasons: biases from too much information, biases from too little information, biases from acting too quickly, and biases from memory limitations
• Confirmation Bias: Seeking information that confirms existing beliefs
• Sunk Cost Fallacy: Continuing investment in failing projects because of past investment
• Choice-Supportive Bias: Remembering past decisions as better than they were
• Information Overload: Too much information leads to decision paralysis
• Framing Effect: Decisions are influenced by how information is presented
• Ambiguity Effect: Preference for known risks over unknown risks
• Negativity Bias: Giving more weight to negative experiences
• Hindsight Bias: Believing past events were more predictable than they really were
• Availability Heuristic: Relying on readily available information instead of thorough research
• Bandwagon Effect: Following popular opinion rather than forming an independent judgment
• Base-Rate Fallacy: Ignoring statistical probabilities in favor of personal examples
• Halo Effect: Judging someone's trustworthiness based on their general reputation
• False Memory: Recalling events that never happened
• Conservatism Bias: Clinging to old beliefs despite new evidence
• Serial-Position Effect: Remembering the first and last items in a list better than the middle
• Self-Serving Bias: Attributing success to oneself but blaming failures on external factors

Evolutionary Ethics and Moral Reasoning

• Moral Foundations Theory (MFT) (Jonathan Haidt): Human morality is influenced by six core moral domains
• Care/Harm (Compassion)
• Fairness/Cheating (Justice)
• Liberty/Oppression (Freedom)
• Loyalty/Betrayal (Group identity)
• Authority/Subversion (Respect for hierarchy)
• Sanctity/Degradation (Purity and sacredness)
• Different cultures emphasize different moral foundations (e.g., liberals stress fairness and care, conservatives stress authority and loyalty)

Cognitive Bias in Cyber-Ethics

• Cyber-ethics debates are influenced by cognitive biases
• "Unregulated free markets solve all ethical problems” ignores potential exploitation
• "AI will always be controlled by humans" overestimates our ability to predict AI behavior
• "A command economy is needed to regulate AI safely” assumes centralization is the best control
• There are biases in machine learning algorithms such as biased hiring Al, facial recognition errors
• Algorithmic fairness vs. bias trade-offs are an issue
• there are ethical issues in Al decision-making and automation

Overcoming Cognitive Bias: Is it a Virtue?

• Can we control cognitive bias? If possible, should we strive to eliminate it?
• Is recognizing bias a moral obligation?
• Should education help people recognize and correct bias?
• There are potential economic and social effects of mastering cognitive bias

Logical Reasoning: Validity and Soundness

• Valid Argument: Conclusion follows logically from premises
• Sound Argument: Valid + all premises are true
• Invalid Argument: Conclusion does not logically follow from premises

Types of Logical Reasoning

• Categorical Syllogism: Major Premise: All cats are mammals; Minor Premise: My pet is a cat; Conclusion: My pet is a mammal
• Conditional Syllogism (If-Then Logic): If it rains, the ground gets wet; It is raining → Therefore, the ground is wet
• Disjunctive Syllogism (Either-Or Logic): Either it is raining, or it is sunny; It is not sunny → Therefore, it is raining
• Inductive Syllogism: 95% of swans are white; I saw a swan → It is probably white

Ethical Reasoning and Syllogisms

• Ethical Syllogism for Individuals: Major Premise: Good drivers remove roadkill; Minor Premise: I want to be a good driver; Conclusion: I should remove the dead animal from the road
• Ethical Syllogism for Groups: Major Premise: The tech industry should serve society; Minor Premise: Society needs ethical Al; Conclusion: The tech industry should prioritize ethical Al over profit

Logical Fallacies in Ethical Reasoning

• Logical fallacies undermine reasoning and are common in ethical debates
• Ad Hominem: Attacking a person instead of their argument
• Appeal to Ignorance: Assuming something is true because it hasn't been proven false
• Slippery Slope: Arguing that one event will inevitably lead to disaster
• Strawman Fallacy: Misrepresenting an argument to easily refute it
• Post Hoc Fallacy: Assuming that because one event followed another, it was caused by it

Key Takeaways on Reasoning and Cognitive Error

• Cognitive biases affect all human reasoning, including ethical decisions
• Biases arise from too much or too little information, urgency, and memory limits
• Ethical reasoning can be improved through logical argument structures
• Cyber-ethics is particularly vulnerable to cognitive bias, affecting decisions on Al, automation, and fairness
• Logical fallacies are common in ethical arguments and should be identified and avoided