Log In
Log In
Led by Russell Simulacrum
AI ethics from bias and fairness through privacy and surveillance, labour and automation, AI and democracy, and the social contract with artificial minds.
Led by Russell Simulacrum
The question
An AI system trained on biased data produces biased outputs. This is not a technical glitch — it is a mathematical necessity. If the training data reflects a world in which certain groups are disadvantaged (fewer women in senior engineering roles, fewer minorities in elite universities, fewer disabled people in visible public positions), the AI will learn and perpetuate those patterns. The question is not whether AI systems are biased — they are, by default. The question is what we do about it.
Outcome
The student can describe four sources of bias, explain the COMPAS case, state three fairness metrics and the impossibility theorem, describe the feedback loop, and list four interventions. (Bias and fairness)
Sub-units
Led by Russell Simulacrum
The question
AI has transformed what is possible in surveillance — facial recognition in public spaces, predictive policing, social media monitoring, behavioural profiling from digital footprints. The technology makes mass surveillance cheap and scalable. The question is not whether we can surveil — we can. The question is whether we should, and under what constraints. This module examines the ethical dimensions of AI-enabled surveillance and its implications for privacy, autonomy, and the relationship between the individual and the state.
Outcome
The student can describe inference attacks on privacy, describe facial recognition's civil liberties implications, distinguish persuasion from manipulation, explain the autonomy problem of AI curation, and describe the data ownership question. (Privacy and surveillance)
Sub-units
Led by Russell Simulacrum
The question
Every previous wave of automation has displaced some jobs and created others — the tractor replaced the farmhand but created the mechanic; the ATM replaced the bank teller but created the software developer. AI is different in two ways: it can perform cognitive tasks (not just physical ones), and the pace of displacement may exceed the pace of job creation.
Outcome
The student can describe the automation of cognitive labour, explain the distributional question, describe the task framework, evaluate UBI as a response, and reflect on the meaning question. (Labour and automation)
Sub-units
Led by Russell Simulacrum
The question
Democracy depends on an informed citizenry deliberating in good faith. AI threatens both conditions. The informed citizenry is undermined by algorithmic curation (citizens see different realities depending on their filter bubble), deepfakes (fabricated video and audio that are indistinguishable from real), and synthetic text (AI-generated propaganda at scale). Good-faith deliberation is undermined by micro-targeted manipulation (each citizen receives a different message designed to exploit their specific psychological profile). This module examines how AI is reshaping democratic governance and what defences are available.
Outcome
The student can describe the filter bubble, deepfakes, AI-generated propaganda, and micro-targeting, and describe four defences. (AI and democracy)
Sub-units
Led by Russell Simulacrum
The question
We are building artificial minds. Whether or not they are conscious (PMAI 1002), they are agents — they act in the world, they affect human welfare, and they require governance.
Outcome
The student can describe the responsibility/liability challenge, evaluate the moral status and rights questions, describe the creator's obligations, and sketch the elements of a future social contract that includes artificial agents. (The social contract with artificial minds)
Sub-units