003 AI Can Get You There Fast... But Where Are You Going?

🎧 Listen to this episode on Apple Podcasts, Spotify, or your favorite platform.

AI has become the golden retriever of modern work.

It wants to help. It wants to fetch the outline, rewrite the email, summarize the research, create the slide deck, polish the strategy, and then cheerfully offer three more versions before we have even decided whether the first one solved the right problem.

That helpfulness feels like a little moment of magic at first, but it is also part of the risk.

In our conversation on Pixels & Priorities, J and I explored the everyday ethics of AI at work. Not the giant global policy conversation, although that matters too, but the practical version many of us are already facing in product management, UX design, customer experience, leadership, and creative work:

When AI helps us move faster, how do we make sure we are still moving in the right direction?

“AI ethics” sounds like a stiff corporate training module we have to take every Fall, but in daily work, ethics is not abstract. It shows up when we present something we do not fully understand. It shows up when we choose not to paste information into a tool without thinking through the data implications. It shows up when AI gives us a confident answer that may be incomplete, biased, or simply wrong.

As we are pressured to drive for acceleration, it's an important reminder that speed does not transfer accountability away from us.

Speed Without Judgment Creates Faster Mistakes

AI is often sold as a productivity tool. Faster output. Faster research summaries. Faster documentation. Faster ideation. Faster everything.

In our product and design work, faster is not automatically better.

“AI can get you somewhere very fast, but it might be the wrong destination.” – J Schuh

A product manager can generate a roadmap faster, but that does not mean the roadmap reflects customer needs.

A UX designer can generate concepts faster, but that does not mean the experience is usable.

A team can summarize research faster, but that does not mean the synthesis is accurate.

A leader can produce a strategy deck faster, but that does not mean the strategy is sound.

Speed is useful only when direction is clear.

AI acceleration means teams can produce more artifacts in less time, but product work is not judged by artifact volume. It is judged by whether we solve meaningful problems for real people.

When AI helps us produce polished work quickly, it can create the illusion of confidence. A well-written document can feel complete even when the thinking underneath it is thin.

Before using AI-generated output in a product, design, or leadership context, pause and ask:

• What problem are we solving?
• Does this output match the original goal?
• What assumptions are hidden inside it?
• What would I need to verify before sharing this?
• Can I explain this clearly if someone challenges it?

AI Makes It Easy to Present Work We Do Not Understand

One of J’s biggest concerns came from his experience as a professor of design thinking and UX. He described students presenting AI-generated slides and then struggling to explain what their own bullet points meant.

That situation is not limited to the classroom.

Many of us have seen some version of this at work. A document sounds polished. A strategy sounds smart. A recommendation sounds authoritative. Then when someone asks a deeper question, a moment of quiet panic washes over the presenter’s face.

Our roles require us to stand behind decisions. We are not just producing words on a page. We are shaping experiences, influencing priorities, guiding teams, and affecting customers.

You have to be able to stand behind what you’re building.

Using AI responsibly means reviewing, understanding, and validating the work before we present it as our own thinking.

Try using this simple review lens before sharing AI-assisted work:

• Comprehension: Do I understand every claim?
• Accuracy: What needs fact-checking or validation?
• Relevance: Does this support the actual goal?
• Accountability: Would I be comfortable defending this in a stakeholder meeting?
• Customer impact: Could this create confusion, harm, exclusion, or mistrust?

Ultimately, we are learning these processes together in real-time. And as we use AI tools, we are in some ways helping build them.

With the rate of change and rising expectations, we are bound to mess up despite our best efforts. Remember to take a breath, and when you learn better, do better.

Golden Retriever Problem: AI Keeps Offering More

“AI is like a happy golden retriever. ‘I can do that for you! And this! And here’s an outline! And here’s another suggestion!’” – Metsy Rose

AI does not simply answer. It expands, suggests, elaborates, and produces options. It invites us to keep going.

That can be incredibly useful for brainstorming, writing, research synthesis, ideation, and planning, but it can also make it harder to draw boundaries.

More content does not always mean better thinking. More suggestions do not always mean better strategy. More automation does not always mean better customer experience.

AI tools are designed to continue assisting, and that helpful momentum can subtly encourage overuse or over-reliance.

Product and design teams need boundaries around what AI should do, what humans must review, and where sensitive or high-impact decisions require extra caution.

Create team-level guardrails around:

• What data can and cannot be entered into AI tools
• What outputs require human review
• If/how AI usage should be disclosed internally
• Which decisions cannot be delegated to AI
• How to validate AI-assisted research, UX, strategy, or prioritization work

Optimization Can Become Manipulation

One of the more unsettling parts of our conversation was the future of AI assistants at work.

J painted a picture of logging into work and being greeted by an AI assistant that knows your schedule, your metrics, your priorities, and your productivity patterns. Helpful, right? Maybe.

Does the AI assistant nudge behavior? Encourage overtime? Track keystrokes? Influence choices?

At what point does helpful become manipulative?

I am concerned these possibilities will be on our doorstep more quickly than we are prepared to deal with. The more AI learns our tone, habits, writing style, and work patterns, the more it can optimize our output.

AI systems can shape behavior, not just support tasks. In product and workplace contexts, that creates ethical questions about autonomy, consent, transparency, and trust.

UX professionals and product leaders will increasingly be asked to design or evaluate AI-driven experiences that influence user behavior. That means ethics cannot sit outside the product conversation.

When designing AI-assisted experiences, we can ask:

• Are users aware they are being nudged?
• Is the recommendation serving the user or only the business?
• Can users opt out?
• Is the system transparent about what data it uses?
• Could this create pressure, dependency, or loss of agency?

Data, Bias, and Accountability Are Product Concerns

AI ethics often becomes clearer the moment we talk about data.

What information is allowed in AI tools? Who can see it? What approvals are required? What happens to proprietary information, customer data, employee information, research transcripts, or strategic documents?

That risk calculation is not theoretical. Many of us are already making those decisions in small ways every day.

The same is true for bias.

In the episode, we talked about AI in hiring, where applicants use AI to optimize resumes while companies start to use AI to screen candidates. J raised the concern directly: are we amplifying bias? Almost certainly.

I want to remind us - bias is embedded in the people building the tools, so it’s logical to assume that the tools are influenced by that human bias.

This reality matters deeply for product management, UX design, and customer experience because the systems we build often inherit the assumptions of the people, data, and organizations behind them.

AI bias happens when AI systems produce unfair, incomplete, or distorted outputs based on biased data, flawed design, or human assumptions baked into the system.

If AI influences hiring, prioritization, customer support, design decisions, or product recommendations, bias can scale quickly and quietly.

Build AI review into product and design workflows:

• Test outputs across different user groups and scenarios
• Review data sources and assumptions
• Include diverse perspectives in evaluation
• Document known limitations
• Treat AI recommendations as inputs, not final answers

Human Judgment Is Becoming More Valuable

AI can generate a mountain of output, but someone still has to decide what matters. Someone has to connect the work back to user needs. Someone has to notice when the answer sounds polished but wrong or whether the thing being built should exist at all.

That someone is still us.

“Our role shifts from creator to curator, validator, and ethical decision-maker.” – J Schuh

And as AI becomes more capable, human judgment may become more important, not less.

Questions for Reflection

1. Where are we currently using AI to move faster, and where might we need stronger review?
2. What types of AI-assisted work should require human validation before being shared or shipped?
3. How do we protect trust with users, customers, teams, and organizations while experimenting with AI?

Key Takeaways

• AI can accelerate work, but it cannot replace human accountability.
• Product managers and UX professionals must be able to explain, defend, and validate AI-assisted work.
• AI guardrails are not bureaucracy. They are how we protect users, organizations, and each other.
• Data privacy, bias, and manipulation are practical product and design concerns.
• The more AI generates, the more valuable human judgment becomes.

Final Thoughts

AI can help us move faster. It can help us learn, create, summarize, explore, and refine. However, it cannot carry our integrity for us.

As product managers, UX professionals, leaders, and creatives, our responsibility is not just to produce more. It is to understand what we are producing, why it matters, who it affects, and what risks come with it.

The future of AI at work will not only be shaped by the tools themselves.

The future will be shaped by the humans willing to ask better questions before pressing send, ship, publish, or approve.

– Metsy
Co-host, Pixels & Priorities

Stay Connected

🎧 Listen to new episodes every Tuesday on Apple Podcasts, Spotify, or your favorite podcast platform. Please leave us a rating, submit a review, and share with a friend. It helps more people find the show and join the conversation.

Connect on LinkedIn: Metsy Rose | J Schuh | Pixels & Priorities