Designing with AI: How UX Teams Can Responsibly and Creatively Use Tools Like ChatGPT

As a designer working at the intersection of complex systems and real human needs, I’ve seen firsthand how AI (when thoughtfully applied), can elevate digital experiences. At organizations like Oracle and RBC, I’ve led initiatives where AI wasn’t just a trend to chase, but a tool we integrated deliberately to solve design problems, enhance efficiency, and deliver better products.

Tools like ChatGPT are rapidly becoming essential companions in the UX process—not just for ideation or copywriting, but for prototyping intelligent experiences, stress-testing flows, and accelerating content iteration. But with this new power comes new responsibility.

AI as a UX Companion: Where It Adds Value

In client projects and internal enterprise tools, I’ve used ChatGPT to simulate user-agent interactions, rapidly iterate on UX microcopy, and model conversational flows. Whether designing complex data tools at Oracle or food safety workflows at Provision, AI helped identify blind spots and reduce friction—especially where UX writers or product strategists were unavailable early in the cycle.

Rather than replacing UX expertise, AI enhances it. ChatGPT offers fast, contextual suggestions that speed up prototyping while leaving the final judgment to human designers. This balance is crucial: the AI is the assistant, not the author.

Real Use Cases in Complex Environments

At RBC, working on enterprise design systems and compliance-heavy interfaces meant rigor was non-negotiable. I used ChatGPT to draft compliant, plain-language microcopy for financial tools and to simulate user dialogues for retirement program flows—always cross-referenced with legal and accessibility standards.

At Oracle, we explored AI-driven onboarding experiences where ChatGPT helped test tone, clarity, and user empathy in automated support content. By prototyping with ChatGPT, we aligned internal stakeholders faster and brought consistency across diverse product lines.

Anonymize Always: Ethical Use of AI in Client Projects

One of the most important habits I’ve built into my process is anonymizing all client content before feeding it into any AI tool. Whether working with large financial institutions or growing SaaS platforms, protecting client data is foundational.

Before using ChatGPT, I scrub real user data, redact identifiable references, and abstract scenarios into generalized patterns. This ensures I respect both privacy and NDAs, while still gaining the strategic benefit of AI collaboration.

Toggle, Don’t Train: Understanding Data Controls

Another key practice is toggling off training settings—or using tools like ChatGPT Enterprise or local LLMs with clear data policies. Many teams worry about “training the AI” with confidential info. The truth is: ChatGPT doesn’t retain or learn from your content unless data sharing is explicitly enabled. But UX designers should always err on the side of caution.

When I introduce AI into client workflows, I make sure stakeholders understand these distinctions. Transparency builds trust and equips teams to make smarter, safer decisions about AI use.

Bridging UX Writing Gaps

A recurring challenge in product development is the missing copy problem, when wireframes are ready, but the words aren’t. AI helps fill that gap, generating first-pass microcopy, onboarding messages, and fallback states quickly and contextually.

Of course, everything AI generates goes through human refinement. But having a draft (something tangible to critique) is far better than designing around “Lorem ipsum.” It lets teams focus on clarity, tone, and hierarchy early, which improves usability and reduces rework later.

Final Thoughts: Responsible Innovation

Using ChatGPT in UX is not about cutting corners, it’s about elevating the design process. It gives designers more headspace to focus on human needs, while handling the tactical lift of early content and interaction modeling.

At its best, AI is a companion that helps us prototype smarter, write faster, and think deeper. But it only works if we treat it responsibly: anonymize inputs, respect data boundaries, and remember that good design is still (always) human-centered.