Stop killing polar bears: AI ethics from the user's side

I've been asked a few times recently whether AI is ethical. It's a fair question, and usually a loaded one. This spring, former Google chief executive Eric Schmidt stood up at the University of Arizona's graduation and was booed. What struck me was that his message was a good one. His point was that the real question isn't how AI will shape our future, but how graduates, and all of us, will shape AI. I agree with him. But the graduates didn't want to hear it from him, and I think I understand why. Being told you're in charge of a technology by one of the people who built it doesn't feel like being in charge.

So let's be honest about the builders. The data centres, the energy and water they use, and the way chips, cloud and software are concentrating in fewer and larger hands are real concerns, and nobody should wave them away. But those decisions will be made by regulators, governments and very large companies. If you run a fifty-person business, your influence there is limited. If we're going to shape AI, it won't be from the top down.

Where you do have influence is in how you and your people use it. I've spent 30 years in technology and watched the internet, e-commerce and social media each arrive the same way. The tool itself was rarely the problem. The habits we formed in the first few years were, and they set in long before the rules caught up.

I'll own my part in this. When AI tools first landed, our family group chat filled up with reworked photos, rewritten messages, videos and a new generation of memes, and I was as guilty as anyone. My daughter had one reply for all of it: "Stop killing polar bears." It was a joke, but she'd understood something the rest of us hadn't. Every question you put to an AI model is broken into units of text, called tokens, and every one of them uses compute, energy and money. It isn't free just because it feels free.

So what does ethical use look like in a business? I'd start with three questions. What is this for? AI is at its best as a reasoning tool, weighing options and finding the flaw in a plan before it costs you money, and at its most wasteful when it's used just because it's there. What does it cost? Vague prompts, endless retries and the most powerful model used for the most trivial job all add up. And who owns the decision? AI can test a decision hard, but there should always be a named person accountable for it, and you should know exactly whose data went in.

Here's the uncomfortable bit for leaders. Very little AI waste is careless. Most of it is untrained. People get handed a powerful tool with no guidance on how to use it well, and when that goes wrong it's a leadership failure, not theirs. I've made that mistake with other technology, and I'd rather people didn't repeat it with this one.

That's why the first thing I put in place with any business is ground rules for AI use, before anyone touches a tool. It's the cheapest point at which an organisation decides what kind of AI user it wants to be. And it matters beyond your own walls. When enough businesses ask hard questions about cost, data and purpose before they buy, providers notice.

Governments will govern, and the builders will keep building. How we use it is up to us. Or, as my daughter would put it, stop killing polar bears.

Source reference: NPR, "Advice for 2026 commencement speakers: don't bring up AI", 20 May 2026.