I didn’t start out caring about AI ethics in the way that phrase usually gets used. Ethics, in most corporate contexts, means compliance — the legal minimum, the box to tick, the policy that lives in a folder nobody reads. I’ve never found that particularly interesting.
What I care about is something more practical: when AI is embedded in real decisions that affect real people, and something goes wrong, what happens? Who knew? Who was watching? Who’s responsible? And — the question I find most interesting — what would it have taken to catch the problem before it mattered?
I got here the practical way. Working with organisations that were deploying AI in customer-facing and operational contexts, I kept seeing the same pattern: fast adoption, thin oversight, and a genuine belief that the model was more reliable than it was. Not malicious — nobody was trying to harm anyone. Just a collective underestimation of how quietly AI systems can fail in ways that are hard to detect and easy to rationalise.
Responsible AI, to me, means building the habits and structures that catch those failures early. It means asking “what could go wrong here, and how would we know?” before deployment, not after. It means treating AI governance not as a compliance function but as a core operational discipline — the same way we treat financial controls or safety protocols in high-stakes environments.
That’s the lens I bring to this section of the blog. Not theoretical ethics. Practical accountability.