The question AI can’t answer for us

Introducing Ethical by Design: Good Technology Principles for AI 

There’s a particular kind of vertigo that comes from watching a technology reshape the world faster than our frameworks can keep up. We’ve been here before with the internet, social media, and the smartphone. Each time, the ethical reckoning arrived late, if it ever came at all. Foreseeable and preventable harms accumulated while the builders moved fast, and the rest of us were left to sort through the wreckage. 

With the proliferation of artificial intelligence, we don’t have the luxury of retrospect. Systems are already making decisions about who gets a loan, who gets flagged at a border, whose résumé clears the first filter. The question of what AI should do, not just what it can do, is no longer theoretical. It’s operational and it’s urgent. 

This is the moment that Ethical by Design: Good Technology Principles for AI is written for. 

Building on what endures 

The Ethics Centre has been asking hard questions about technology and human flourishing for a long time. When we published the original Ethical by Design principles in 2018, AI was largely a specialist concern: something debated in research labs and tech conferences, not yet woven into the fabric of daily life. Those principles – that technology must respect human dignity, anticipate harm, and serve a genuine purpose – were designed to be durable. And they are: the ethical foundations haven’t shifted. 

But the landscape has. 

Generative AI didn’t just accelerate existing trends; it introduced a qualitatively different kind of challenge. Systems that learn, adapt, and produce outputs their own designers do not intend and cannot fully explain don’t fit neatly into frameworks built for more legible technologies. When a model generates a deepfake, hallucinates a legal precedent, or encodes a historical bias into a hiring recommendation, the question of accountability doesn’t resolve cleanly. The complexity is real, and it demands a response equal to it. 

This updated framework extends the original Ethical by Design principles. What’s new is the application. The framework intentionally grapples seriously with dimensions of AI ethics that are newer to mainstream conversation: the environmental cost of training and deploying models at scale; the hidden labour of the data annotators and content moderators whose work makes AI possible; the specific risks of synthetic content in an information environment already struggling with trust. These aren’t edge cases. They’re central to what it means to build AI responsibly. 

For the people who build and lead 

If you’re making decisions about whether and how to deploy AI in your organisation, this framework is for you. Not as a compliance checklist, but as a set of principles rigorous enough to stress-test your own reasoning. The central question of this framework is deceptively simple: ought before can.

Before we ask whether an AI system is technically feasible or commercially viable, we ask whether it should exist: whether its purpose is legitimate, its means are ethical, and its costs are fairly distributed.

That question doesn’t have a technical answer. No model can generate it for you. It requires the kind of careful, honest ethical reasoning that The Ethics Centre exists to support. This framework is designed to make that more accessible. 

Commercial pressures and competitive dyanmics are real. The expectation that organisations will adopt AI quickly, confidently, and at scale is entirely real. None of that changes the ethical calculus. It just makes the discipline harder and more necessary. 

What we’ve tried to do here is give you something useful for that harder work: eight principles with philosophical grounding and concrete practices, applied to the specific terrain of AI systems. Principles that don’t pretend the choices are easy, but insist that they are choices that can and should be made with explicit intention and accountability for impact. 

The revolution is already here. The question is what we choose to build with it, and who we choose to protect in the process. 

Download a copy of Ethical by Design: Good Technology Principles for AI here. 

If you’d like to discuss implementing ethical AI in your organisation, contact consulting@ethics.org.au.