COMP90087 · The Ethics of Artificial Intelligence
Introduction to AI Ethics and Principles
Week 2 hands you the toolkit the rest of the subject uses: the ethics-versus-law distinction, principlism (beneficence, non-maleficence, justice, respect for autonomy), Australia’s AI Ethics Principles, and Bietti’s ethics-washing vs ethics-bashing critique. Principlism is named explicitly in the in-class essay prompt, so this is directly assessed — you must be able to apply the four principles to weigh an AI deployment’s harms and benefits. Tasioulas’ “humanistic ethics” (the three Ps) is the high-value compulsory reading behind the exam’s deeper questions.
What this chapter covers
- 01Principlism — the four bioethics principles applied to AI: beneficence, non-maleficence, justice, respect for autonomy (named in the essay prompt)
- 02Australia’s AI Ethics Principles (voluntary set: human wellbeing, human-centred values, fairness, privacy, reliability/safety, transparency, contestability, accountability)
- 03Ethics washing (using ethics language to pre-empt regulation) vs ethics bashing (dismissing moral philosophy as useless) — Bietti
- 04Weighing harms and benefits of an AI system — the core skill of the in-class essay
- 05Tasioulas’ two core questions of ethics: what is a good/flourishing life, and what do we owe others?
- 06Tasioulas’ critique of preference-based utilitarianism / the “optimising mindset”
- 07The three Ps of humanistic ethics — pluralism (incommensurable values), procedures not only outcomes, participation
- 08Human dignity as inherent equal worth; a focus on narrow AI rather than near-term AGI
Apply principlism (the four principles) to an AI deployment
- +1Beneficence (do good). The system could genuinely reduce dumping, keep public spaces clean and free up ranger time — a real public good if it works as intended. Name the benefit concretely rather than asserting it.
- +1Non-maleficence (avoid harm). Weigh false positives (wrongly fining an innocent resident), chilling surveillance of a whole neighbourhood, and error modes that are hard to detect. The harm is concentrated on those wrongly flagged.
- +1Justice (fairness). Ask whether the cameras are sited disproportionately in lower-income areas, and whether the burden of surveillance and fines falls unequally — a disparate-impact concern even if the rule is “the same for everyone.”
- +1Respect for autonomy. Can residents consent to being watched, contest a fine, and get an intelligible reason for it? An automated penalty with no route to challenge treats people as objects of enforcement rather than as agents — land a position such as “deploy only with a human-reviewed appeals process and transparent siting.”
Key terms
- Principlism
- Applying four classic bioethics principles — beneficence (do good), non-maleficence (avoid harm), justice (fairness) and respect for autonomy — to AI decisions. Named explicitly in the in-class essay prompt.
- Beneficence / non-maleficence
- The paired duties to actively do good and to avoid causing harm. In AI cases the tension is usually a real benefit (accuracy, efficiency) set against undetected or concentrated harms.
- Ethics washing
- Using the language of ethics superficially to create an appearance of responsibility and pre-empt or avoid binding regulation (Bietti). Its over-correction, ethics bashing, dismisses moral philosophy itself as useless.
- Australia’s AI Ethics Principles
- A voluntary set issued by the Department of Industry — including human wellbeing, human-centred values, fairness, privacy protection, reliability and safety, transparency and explainability, contestability and accountability — that students read as the applied-principle canon.
- The three Ps (Tasioulas)
- Humanistic ethics rests on Pluralism (many incommensurable values, so no single optimising function), Procedures (how a decision is reached matters, not only the outcome) and Participation (well-being needs active engagement, not passive end-states).
- Value incommensurability
- Tasioulas’ claim that plural values (justice, friendship, achievement, understanding) cannot all be reduced to one currency or “master concept” — so ethics cannot be a single optimisation.
Introduction to AI Ethics and Principles FAQ
What exactly is ‘principlism’ and why does it matter for assessment?
Principlism is the practice of applying four bioethics principles — beneficence, non-maleficence, justice and respect for autonomy — to an AI decision. It matters because the in-class essay prompt names it explicitly, so you may be asked to defend a position on an AI scenario by trading these four principles against each other. Practise applying all four to a fresh case and reaching a conditional verdict.
What is the difference between ethics washing and ethics bashing?
Both are failures Bietti warns against. Ethics washing uses ethics language superficially to look responsible and head off regulation; ethics bashing over-corrects by dismissing moral philosophy as useless. The “view from within moral philosophy” defends serious ethical reasoning against both — a neat MCQ pairing and a good essay nuance.
Do I have to memorise all of Australia’s AI Ethics Principles?
You should recognise the set and be able to name several (fairness, privacy protection, transparency and explainability, contestability, accountability, reliability and safety, human-centred values, human wellbeing) and use them as a checklist when weighing a deployment. Being able to apply them to a case is worth more than rote listing; confirm the current list on Canvas.
Can AI help me apply principlism to a case in COMP90087?
Yes, as a tutor. Sia is built to mirror how COMP90087 is assessed at the University of Melbourne: give it an AI scenario and it will walk you through beneficence, non-maleficence, justice and autonomy step by step and check whether your verdict actually follows from your reasons. It explains the method and does not write your graded essay — University of Melbourne academic-integrity rules and the subject’s GenAI policy apply.
Exam move
Treat the four principles as a permanent checklist and rehearse them on one fresh AI scenario a week, because the in-class essay explicitly asks you to apply principlism and trade the principles against each other. Make a two-column sheet: principle → the concrete question it forces (beneficence = what real good? non-maleficence = who is harmed and how invisibly? justice = does the burden fall unequally? autonomy = can they consent and contest?). Learn Bietti’s washing/bashing pair as a crisp MCQ, and read Tasioulas actively — the three Ps (pluralism, procedures, participation) are the reading most likely to power a deeper short answer. Weaving these into your revision now protects both essays and the closed-book hurdle exam.
Working through Introduction to AI Ethics and Principles in COMP90087? Sia is AskSia’s AI AI Ethics tutor — ask any COMP90087 Introduction to AI Ethics and Principles question and get a clear, step-by-step explanation grounded in how COMP90087 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
