TL;DR
Ethical concerns about AI (water use, jobs, copyright) are real and worth taking seriously. But the right response isn’t to disengage. If the people who care most about ethics opt out, AI evolves without them. And that’s the version of AI none of us want.
Table of Contents
What is AI ethics?
Before getting into the instinct to opt out, it is worth naming what we mean by AI ethics here: the responsible design, deployment and use of artificial intelligence, with attention to fairness, transparency, accountability and sustainability.
Why AI ethics often leads people to opt out
There is a pattern I’ve been seeing in Copilot enablement training session feedback lately.
The session had gone well. People were engaged, the demos landed, and we finished on time, which, if you’ve ever run a Copilot enablement session, feels like its own small win. Then as I read through the responses, there would be one or two that said something along the lines of: “I’m still not convinced AI is ethical. What about the water usage?”
Here’s the thing. They hadn’t attended. They’d already decided AI was bad, so they didn’t come, and then they filled out the training feedback form anyway.
It’s a pattern. The concern is real enough to act on, real enough to skip the session, real enough to write about it, but the action it leads to is disengagement. Staying outside the room. And I think that’s the wrong call.
I’ve started to call it the Water Moment: the point where someone’s ethical instincts about AI pull them toward opting out rather than leaning in. It doesn’t always look like an empty seat. Sometimes it’s a quiet comment in a one-on-one, or a question that surfaces at the end of the session.
What are the main concerns in AI ethics?
The concerns are real. Start there. The worst thing you can do when someone raises an ethical concern about AI is deflect. It signals that you either haven’t thought about it or that you’re protecting the technology instead of engaging with it honestly. Neither is a good look.
AI infrastructure does consume significant energy and water. Data centres require cooling, and cooling requires electricity and, in many cases, water. Microsoft and other major cloud providers have published data on this, and the numbers are large enough to take seriously. On jobs, automation does redistribute work, and it’s reasonable to ask whether organisations are being thoughtful about that redistribution. On copyright, AI models were trained on vast datasets of human-created content, and the legal and ethical questions around that are genuinely unresolved. There is no tidy answer here that you can hand someone to make the discomfort go away.
The people raising these concerns aren’t being difficult. They’re paying attention. And they deserve a response that takes them seriously.
Does avoiding AI reduce its ethical impact?
Every search, every email, every document we work on is already running through data centres that need to be powered and cooled. AI increases the intensity of that demand, but it didn’t create the system. And the levers to address it sit well above the level of any individual prompt.
And that’s where I want to push back on the conclusion that feels ethical but I don’t think quite holds, which is: therefore, I won’t use AI.
Here’s what that argument misses. The infrastructure isn’t something your individual choice touches either way, it’s already there, already running, already being built out with or without your participation. So the question isn’t really about the infrastructure at all. The more honest question is what your absence actually changes, and whether opting out is a decision you made, or one you defaulted to.
Is opting out of AI an effective ethical response?
I get it. When something has a cost, opting out feels like the clean response. It aligns with a precaution mindset, with the instinct not to be complicit in something you’re not sure about.
But the infrastructure is already in place. The models are already built. And the picture on the two most-cited concerns is more mixed than a first glance suggests. On the environmental side, providers like Microsoft now use recycled and reclaimed water at several of their datacentre sites to reduce reliance on drinking water, and individual decisions to skip tools like Copilot are unlikely to materially shift overall energy demand either way. On copyright, lawsuits, settlements, and licensing deals with publishers are already reshaping how AI companies compensate creators. None of that erases the concerns, but it complicates the idea that staying out is the more responsible choice. The levers that actually move the dial sit elsewhere, with the people who design, build, and govern this infrastructure, not the everyday users typing the prompts.
The people who show up to shape how these tools are used, that’s where the real influence sits. If the people who care most about ethics, who ask the best questions, who push back when something isn’t right, decide that their response is to sit this one out, then the rollout in their organisation doesn’t pause. It continues, with one fewer voice in the room asking whether it’s being done well.
Who is responsible for ethical AI use?
When someone asks me about AI and water, I try to map it out honestly.
- Energy sourcing sits with infrastructure providers.
- Data centre design and cooling choices sit with vendors.
- Workforce impact sits with organisations and their leaders.
- Copyright and licensing choices sit with model developers, vendors and regulators.
- The individual end-user, the person in the training room writing a summary or drafting a document, is at the bottom of that chain.
That doesn’t mean individual behaviour doesn’t matter at all. Intentional use matters. Using AI for genuine productivity rather than novelty use matters. Critical thinking about outputs matters. Not passing AI-generated work off as purely your own, especially in contexts where that distinction is ethically significant, matters.
You cannot fight something you don’t understand, and you cannot shape something you’ve never touched. The people best positioned to push back on how AI gets rolled out in their organisations, to ask the right questions in procurement conversations, to advocate for responsible governance, are not the ones who opted out. They’re the ones who actually sat with the tools, saw where they work, noticed where they fall short, and built a real picture of the trade-offs from experience rather than assumption. That’s the level of understanding you need to have a seat at the table when the governance conversations happen. And you don’t get there by staying outside the room.
What can AI ethics learn from cloud adoption?
This pattern isn’t new. Cloud adoption went through almost exactly the same arc: security concerns, loss of control, fear of change, resistance from the most thoughtful people in the room, and then, slowly, structured enterprise adoption with governance frameworks built along the way.
The concerns about cloud weren’t wrong either. There are real trade-offs in moving infrastructure off-premises. What changed over time wasn’t that the concerns disappeared. It was that they got folded into how organisations approached the technology through:
- Procurement criteria
- Security requirements
- Compliance frameworks
- Vendor accountability
AI is following that same path, and it needs the same thing: people who care enough to stay engaged and push for governance, not people who care so much that they leave.
And some organisations aren’t waiting for regulation to catch up. Companies are already building sustainability criteria into how they procure and deploy AI, setting internal standards around energy use, vendor transparency, and responsible sourcing. That kind of pressure, coming from inside organisations rather than just from policy, is exactly what moves the dial. But it only happens when the people who care are in the room to push for it.
How can you apply AI ethics in practice?
If you care about AI sustainability, the most useful thing isn’t to opt out. Stay engaged and use that concern as fuel for action.
Learn the tools properly. Not just the features, but the trade-offs. Understand what AI can and can’t do, where it falls short, and what the real costs are. You can’t advocate for responsible use if you’ve never used it.
Use your concern as a lens, not a reason to leave. If you care about the environment, copyright, labour impact or governance, that concern can make you a sharper practitioner. It helps you ask whether AI is the right tool for the task, whether the compute cost is proportionate, and whether an agentic workflow is genuinely useful or just exciting because everyone else is talking about it. So don’t leave your values at the door. Bring them in and put them to work.
Know what your vendor is committing to. What are their sustainability commitments? How do they measure and report on energy and water use? Microsoft’s 2025 Responsible AI Transparency Report and sustainability commitments are one example of what published accountability looks like. If you want to understand what’s actually being measured and why it matters, the Environmental Law Institute’s breakdown of data centre water use is worth reading first.
Push internally for responsible procurement criteria. Energy sourcing, vendor transparency, and responsible sourcing should be part of how your organisation evaluates and deploys AI tools, not an afterthought.
Staying in the room is how you make it better
Responsible AI means building and using AI systems that are fair, transparent, accountable, and aligned with human values, and that requires participation from people who actually hold those values.
The instinct to protect yourself from complicity is understandable, genuinely. But opting out doesn’t change what happens in your organisation. It just removes one thoughtful voice from the decisions that shape how these tools are used in your team, your workflows and your organisation.
If your organisation is rolling out AI right now, your voice carries more weight than you might think. The strongest position is not blanket acceptance or blanket refusal. It is informed participation: using the tools, seeing the trade-offs clearly, and being willing to challenge the way they are adopted, especially when no one else in the room is.
So the next time someone in your session raises their hand and asks about the water, thank them. They’re exactly the kind of person AI adoption needs more of. And then give them a reason to stay.
Frequently asked questions
Is using AI ethical?
It can be, but it depends on how it’s used and governed. Ethical AI is less about the technology itself and more about whether it’s applied responsibly, with attention to transparency, fairness, and accountability. And in most organisations, the ethical outcome really does come down to how people choose to use and shape these tools, not the tool itself.
Does AI increase environmental impact?
Yes, AI workloads can increase energy and water usage, particularly in large-scale data centres. But those impacts are driven mostly by infrastructure design, energy sourcing, and vendor decisions, not by individual end users typing prompts. Which is why broader organisational and industry-level action tends to move the needle far more than individual usage choices ever will.
Should I stop using AI for ethical reasons?
Opting out might feel like the ethical choice. But in practice, it mostly limits your ability to influence how AI gets used, since organisations will keep adopting it either way. The more effective move is staying engaged, understanding the trade-offs, and helping shape responsible use and governance from within.
How can I use AI responsibly?
Focus on intentional, thoughtful use: applying critical thinking to outputs, being transparent about where AI shows up in your work, and questioning whether it’s even the right tool for the task. But the biggest impact goes beyond individual usage, it comes from advocating for responsible governance, vendor transparency, and sustainability within your organisation.
