Before any debate about how to govern AI or handle big data, there is a prior question: is there anyone there to be wronged? These notes start where the ethics actually starts, with the philosophy of mind. They ask what consciousness is, whether it depends on biological material or only on functional organization, and whether the systems built in the deep-learning era could ever cross from doing into feeling. The governance and surveillance questions in AI governance and surveillance and privacy assume an answer to this one. The wing makes that assumption explicit.
Foundations: What Consciousness Is
The distinctions that make the rest of the debate possible to reason about.
- Consciousness: Access vs Phenomenal - the functional “available for use” sense versus the felt “what it is like” sense, and why moral status hangs on the second
- Functionalism and Multiple Realizability - mental states defined by what they do, not what they are made of, and the octopus argument that opens the door to non-biological minds
- Physicalism and the Mind - why a physical mind removes the soul barrier to machine consciousness without saying which systems are conscious
- Scientific Theories of Consciousness - global workspace, higher-order, and integrated information theory as rival markers for when feeling is present
The AI-Consciousness Debate
The live argument over whether today’s systems have, or could be built to have, whatever consciousness requires.
- Could an LLM Be Conscious? - Chalmers’s six objections, most framed as missing features rather than permanent barriers, and the 25%-or-more credence for a conscious successor system
- The Biological-Substrate Objection - Block’s case that electrochemical “meat” may be necessary, and the extrapolation dilemma that pulls AI and insect consciousness apart
Responsibility and Moral Agency
Once systems act in the world, who answers for what they do, and can a machine itself be a moral agent.
- The Responsibility Gap - autonomous weapons and the worry that harm can occur with no one who had enough control to be justly blamed, and the optimist reply
- Can AI Be a Moral Agent? - mind-less morality (accountability without responsibility) against the claim that sentience is required, and algorithms as “moral zombies”
Fairness and Big Data
Why fairness criteria conflict, where the conflict is really a values choice, and what the categories themselves hide.
- The Impossibility of Algorithmic Fairness - COMPAS, and why calibration and equal error rates cannot both hold when base rates differ
- Fairness as Equal Concern - the attitudinal turn: unfairness as unequal concern living in a utility function, no mind required
- Social Categories and Machine Learning - why the protected attributes ML optimizes over may not be things you can coherently hold fixed
Moral Advice and Self-Knowledge
Should we let AI guide our moral decisions, and what it costs the agent when we do.
- AI as Moral Advisor - the artificial moral advisor, the asymmetry of moral deference, and the “digital ethical twin” that trains on your own behavior
Still to come
Notes forthcoming from the remaining threads: privacy and the data economy, plus the adjacent philosophy-of-mind gaps (the Chinese Room, moral patiency, the Turing test).
The full file listing follows below, generated automatically by Quartz.