Working with AI: Toward an Evolving Policy for Design

I started this article more than a year ago. The goal then was to articulate a clear position on the role of AI in our design practice. A kind of stake in the ground, outlining how and when we’d use AI tools, if at all.

Since then, I’ve stopped and restarted an untold number of times. Every time I think I’ve formed a position, something shifts. New tools emerge, new risks are uncovered, and new questions surface. I could easily wait another year to finish this and still feel the same.

It’s tough to define a policy for something evolving as fast and unpredictably as AI. The topic is too vast and unresolved. It touches everything from copyright law to climate impact, from creativity to ethics. Will AI prove to be the transformational force that tech giants and investors are banking on? Or will it become just another tool in the long arc of innovation, offering as many cons as pros?

It’s important to note that these thoughts represent just one perspective among many at Q30, where different voices will delve deeper into these various issues.

For now, we’re trying to engage with it actively and critically. We’re considering it in two key areas: research tools and creation tools.

AI in practice: tools that steer, not replace

At this point, AI isn’t replacing design as we practice it. But it is starting to reshape the workflow around it. Most of the tools we use today aren’t about final output. They’re steering mechanisms, accelerators, or time-savers. They help us move faster, prototype more, explore options, and generate starting points.

That’s useful. But it comes with a caveat: something is lost when you don’t do the work yourself. The very act of compiling workshop feedback or analyzing research data is part of the discovery process. It forces you to sit with the ambiguity, to notice patterns, to understand the problem at a deeper level.

If a machine does all that synthesis for you, what have you really learned?

 

Writing is thinking. To write well is to think clearly. That’s why it’s so hard.

—David McCullough

 

AI writing tools are a good analogy. You can use them to sound polished, but they won’t make you a clearer thinker. The act of writing helps us clarify and refine our ideas. Without that struggle, your voice starts to dissolve into the algorithm’s.

And that’s one of the quiet risks: the erosion of originality. When the machine takes over the bulk of the work, it becomes harder to distinguish where your ideas end and the tool’s output begins.

Design as curation, not invention

To be fair, most of the design is not about radical invention. It’s about selecting the right elements that already exist in the world and applying them to solve the problem at hand. AI won’t generate new colours outside the perceivable light spectrum. It won’t invent a layout that no one has ever tried. It won’t write a word that doesn’t already carry meaning. It won’t create a story whose core elements aren’t already universal.

Like music, design works within constraints. There are only so many notes, tones, and structures, but endless combinations. That’s where the originality lies: not in the parts, but in how they’re assembled. In that sense, AI is a remixer, not a composer. AI is only as good as the person who prompts it, edits it, and curates the results. It can elevate imagination, but it can’t replace it.

The temptation to outsource thinking

We’re clearly in an experimental phase. There’s an avalanche of new tools right now. Like the early days of smartphone apps, we’re watching a flood of innovation that will eventually settle into a handful of ubiquitous platforms. But nobody knows which ones.

The challenge isn’t just keeping up. It’s maintaining a critical stance in the face of ease and automation. We need to ask:

• Is this tool doing the thinking for me?
• Is it helping me explore, or just auto-completing my instincts?
• Do I still understand what I’m doing, and why?

The early shape of a policy: questions we’re asking

As we begin shaping our own internal policy, we’re not looking for blanket rules. We’re looking for guidance. Here are some of the questions we’re asking:

• Am I taking away someone’s job? Especially for things like stock imagery, where creators already earn pennies for their work, are we further commodifying creativity?
• What’s the environmental cost? Training and running large language models consume significant amounts of electricity. Even a single AI-generated image or ChatGPT output has a higher energy cost than we often consider.
• Can I use the output safely? Copyright laws around AI-generated content are still evolving. If I use an AI-generated image in a branding campaign, can I really own it? Can someone else generate a near-identical image using similar prompts?
• Is the output original enough? How different is it from what anyone else could generate using a similar set of prompts?
• Does this make sense for the task at hand? For ephemeral uses, like social media posts, AI-generated images or headlines may be fine. But for brand-defining work? Probably not.
• How do we feel about faking humanity? Using realistic photos of people who don’t exist still feels unsettling. It’s not the same as illustration or stock photography. I’m not sure exactly why, but it’s worth examining. Maybe it’s the erasure of real humans who might have once done that job, or the possibility of creating a likeness so similar to a real person that the distinction becomes negligible. Regardless, this issue extends beyond a mere visual choice. Either way, it’s more than just a visual choice; it’s a philosophical one.

We’re not alone in wrestling with this. One of the top concerns cited in recent surveys is not technical, but human: how do we assess the quality of AI’s output? That’s followed closely by concerns about data privacy and IP protection.

Pragmatic adoption: use it when it makes sense

In the short term, AI tools will be utilized at the discretion of individuals rather than implemented through top-down policies. For example, a researcher might use AI to analyze and categorize data collected from in-person interviews. A developer could rely on AI for coding assistance, while a designer might use it to explore visual metaphors that align more closely with strategy. Additionally, account managers may employ AI to gather information about a client and craft a well-tailored email.

It’s about knowing what tool to reach for, and when.

We don’t need rigid rules. We need critical awareness. We also need to maintain the elements that make design meaningful: curiosity, human insight, intentionality, and craft.

Where we land, for now

We will continue to utilize AI tools when they aid us in exploring new ideas, generating quicker iterations, or uncovering insights more rapidly. However, we will set boundaries where these tools may risk replacing human understanding, voice, and originality.

This is not a final policy; it is a dynamic one that we will review frequently. The technology will improve – or it might not! What is less specific is how our roles as designers will evolve if we rely too much on external thinking.

For now, we will continue to ask questions and do the work.