Bring in the philosophers: why AI ethics needs more than engineers
Sound AI governance depends on questions engineering alone cannot answer, which is why ethics belongs at the centre of certification.
5 min read · IAIDL
Engineers build AI systems. They are not, by training, equipped to decide whether those systems should be built, how they should weigh competing values, or what fairness means in a contested case. Those are questions of ethics, and they need people trained to ask them well.
In 2021 IAIDL was selected for the UNDP initiative Bring in the Philosophers, a programme on AI ethics. The name carries the argument. Getting AI right is not only a technical problem. It is a question of human values, and answering it requires the disciplines that have studied values for centuries.
The temptation in industry is to treat ethics as a compliance checkbox: a policy document filed away, a training module clicked through. That approach fails because the hard cases are not covered by rules written in advance. They require judgement, and judgement has to be developed.
Consider a typical decision. An AI model improves efficiency but reduces transparency. It performs well on average but less well for a minority of users. It is legal but arguably unfair. No engineering metric resolves these tensions. They are value trade-offs, and naming them clearly is itself a skill.
This is why ethics belongs inside competency, not beside it. A professional certified to work with AI should be able to recognise an ethical question when it appears, articulate what is at stake, and reason through it. That capability is as practical as any technical skill.
Philosophy contributes specific tools here. It teaches how to define terms precisely, how to test an argument for consistency, and how to weigh competing principles without collapsing into either rigid rules or pure intuition. These are working methods, not abstractions.
Bringing philosophers into AI does not sideline engineers. It complements them. The engineer asks whether a system can be built and how. The ethicist asks whether it should be, for whom, and at what cost. Good governance needs both questions answered together.
For organisations, the practical risk of ignoring ethics is concrete. Systems that treat groups unfairly, that cannot explain their decisions, or that erode trust create legal, reputational and operational exposure. Ethical competence is a form of risk management, not a luxury.
International standards increasingly reflect this. ISO/IEC 42001:2023, the standard underpinning IAIDL's AIMA maturity assessment, treats responsible governance as integral to managing AI well. Ethics is being written into the frameworks organisations are expected to meet.
A vendor-neutral certification body is well placed to carry ethical competence because it has no product to defend. It can ask whether a use of AI is sound without an interest in selling the tool that performs it. Neutrality and ethical scrutiny reinforce each other.
The broader point is cultural. Treating ethics as central, not peripheral, changes how an organisation approaches AI from the start. Questions get asked earlier, designs get challenged sooner, and problems surface while they are still cheap to fix.
Bringing in the philosophers is shorthand for a serious idea. AI is too consequential to leave its hardest questions to any single discipline. Competence in AI now includes competence in judgement, and that is something worth certifying.