If Predictions and Insights is the category Product Owners skip because it feels uncertain, Ethical Considerations is the category people skip because it feels like someone else’s job. Legal’s job. IT’s job. Leadership’s job. Anyone’s job but the Product Owner sitting in the room deciding what to build next.
That instinct is understandable, and it’s also worth pushing back on a little — because a lot of the actual ethical exposure in AI-assisted product work shows up first at the level of daily decisions a Product Owner makes, long before it reaches legal or leadership.
Where This Actually Shows Up
You don’t need a formal AI ethics committee to run into this. It shows up in smaller, more immediate questions:
Whose data is this, really? When you feed customer feedback, support tickets, or usage data into an AI tool to summarize or analyze, have you actually confirmed that’s an appropriate use of that data — or did it just seem convenient in the moment?
Who checks the output before it becomes a decision? An AI-generated prioritization suggestion, sentiment analysis, or roadmap scenario is only as trustworthy as the human review that happens after it — and it’s easy to skip that review when the output looks reasonable and you’re moving fast.
What happens when the AI is confidently wrong? AI tools don’t hedge the way a cautious colleague would. A wrong prediction or a biased pattern read can come back sounding just as authoritative as a correct one — which puts more weight on you to notice when something doesn’t quite add up.
Building a Simple Framework, Not a Policy Document
You don’t need to write an official ethics policy to make real progress here. A lightweight, personal framework goes a long way:
Before using AI on something, ask: would I be comfortable explaining this use out loud to the person whose data it is? If the honest answer is “I’d rather not have that conversation,” that’s worth noticing before you proceed, not after.
Decide in advance what gets human review, and actually stick to it. Not everything needs the same scrutiny — a first draft of internal notes is different from a customer-facing recommendation. But deciding the line before you’re in a hurry matters more than deciding it after something’s already gone out.
When something feels off, trust that instinct enough to slow down. This sounds obvious, but the whole point of AI output looking polished and confident is that it’s easy to talk yourself out of a hesitation that turns out to be correct.
This Connects Directly Back to Trust
If you’ve read the post on the hardest organizational problems, you’ll recognize this territory — building genuine stakeholder trust in AI-assisted work depends on having a real answer for “how do you make sure this is being used responsibly,” not just “how do you make sure the output is good.” Ethical Considerations and Trust aren’t really two separate problems. One is largely what builds the other.
Try This This Week: The Explanations Test
This week, before you use AI on something that touches someone else’s data or a decision that affects them, run the explanation test: imagine explaining exactly how you used AI here to the person most affected by the outcome. If that explanation feels comfortable and honest, proceed. If it doesn’t, that discomfort is worth listening to before you move forward, not after.
Where This Fits
This is deliberately one of the less comfortable posts in this series, and that’s on purpose — the categories people avoid are usually avoided for a reason, and naming the reason is the first step toward actually engaging with it. If this raised more questions than it answered, that’s a genuinely good thing to bring to the next Alumni Meetup — this is exactly the kind of topic that benefits from more than one perspective in the room.
Keep going:
- Ready to go deeper? Check out our upcoming AI courses for hands-on classes with real practice, not just reading.
- Want more like this? Subscribe to our YouTube channel for walkthroughs and demos that pair well with these posts.
- Have questions or want to talk it through? Reach us directly at hello@rocketninesolutions.com.
- Know a colleague wrestling with the same challenge? Send them this post — half the value of this series comes from the people who’ve already tried it sharing what worked.
A Note on How This Was Written: This post was developed collaboratively between Rocket Nine’s training team and Claude, Anthropic’s AI assistant — drawing on real material from our AI for Product Owners class and anonymized insights from the Product Owners who’ve taken it. We think that’s fitting: the same thoughtful human-AI collaboration we teach in class is exactly how we approached writing about it.
© 2026 Rocket Nine Solutions Inc. All rights reserved.