LECTURES (SLIDES)
What does the “Algocrat” thought experiment shows us?
The "Algocrat" thought experiment illustrates the potential ethical implications of algorithmic decision-making in governance, highlighting concerns about bias, accountability, and transparency in automated systems.
Cite the problem with this definition of AI: “Artifacts (i.e., human inventions) that engage in tasks that normally require
human intelligence.
Many intelligent non-human tasks (communication among birds,
game-playing among dolphins, web-spinning by spiders, etc.). Too narrow!
Cite the problem with this definition of AI: Any complex cognitive activity undertaken by an artifact
Classifies many computers, robots, and calculators as AI—including
thermostats, remote-controlled cars, and software applications. Too broad!
Cite the problem with this definition of AI: “Forms of reasoning (including causal reasoning, decision-making, learning, and perceiving) undertaken by artifacts with at least some degree of autonomy (i.e., independence from explicit programming by humans)
Includes various forms of machine learning, as well as brain simulation and
substitution.
Describe AI today
Currently dominated by methods of machine learning, where algorithms draw inferences from large data sets. “Software that learns by example.”
Explain the 4 types of AI today
Narrow AI: performs specific tasks (e.g., weather prediction, spam filtering, credit decisions autonomous driving)
Multi-purpose AI: performs various content generation and analytical tasks, often in response to natural language prompts (GPT4, Gemini, Claude, LLaMa)
General AI: human-level intelligence that learns without training and can choose its own goals (Samantha from Her, Data from Star Trek)
Superintelligence: exceeds human- level intelligence in all domains (e.g., HAL 9000, The Terminator)
Common misconceptions about ethics
• “Ethics refers to the socially-accepted standards of behavior prevalent
among members of a community.”
• “Ethics is a matter of subjective opinion; there is no objective truth
regarding right, wrong, good, bad, etc.” (cf. Blackman 2022).
• “Ethics depends on religion.”
• “Ethics is complementary to law and policy; it governs us in situations
where there are no formal rules.”
• “Ethics is a barrier to innovation, progress, and excellence. It is for
suckers, fools, and weaklings.”
Define ethics
Ethics is the attempt to identify, and guide our conduct by, the best reasons for acting.
The main question: What do I have most reason to do?
It is a process of reasoning, argument, and justification
5 aspects of ethics
Non-relativistic. The best reasons for action may or may not align with what is
popular or socially accepted.
• Objective. While we will often disagree on what the best reasons are, ethical
positions are open to scrutiny, and some positions are better supported by reasons
than others (Blackman 2022).
• Secular. While many religions prescribe ethical principles and provide foundations
for them, ethical reasoning does not require religious commitments.
• Comprehensive. Ethics not only complements law and policy but also provides the
standards for making and evaluating law and policy.
• Unavoidable. Claiming that we should prioritize (e.g.) innovation, progress, or
excellence instead of some other value(s) is itself an ethical argument.
Define AI governance
he processes, structures, and institutions that exercise
authority and control over AI
• Government regulation
• International law
• Organizational processes and structures
• Industry and professional standards
• Civil society and social norms