Why the ethics of AI-generated design depend on context, intention, and consequence
I recently saw a missing-dog poster that was clearly made with AI. We all know the style by now. The overly polished image, overly energetic typography, and visual choices that somehow announce that a machine was involved. It got me thinking about how some of the more vocal designers opposed to AI would react to something like this.
My question for these folks is simple. Would you criticize someone for using AI to help find their beloved pet? Unless you are a sociopath, I suspect most designers would have no problem with it. Someone needed a functional piece of communication, lacked the ability or resources to create it themselves, and used AI to solve the problem.
Okay, let’s try another example. A local pizza shop that has been around for as long as anyone can remember is having financial difficulty. It cannot compete with the large chain restaurants moving into the area. The owners want to promote the business, but they cannot afford to hire a professional designer, so they use AI to create flyers, social media posts, and other promotional materials.
How about now? I imagine most would still be relatively comfortable with the use of AI here. Yes, a professional designer could have been paid to produce the work, but that assumes the pizza shop had the money to hire one in the first place.
AI did not necessarily take work away from a designer. It may have enabled design work that otherwise would not have existed. This is not entirely hypothetical. A 2025 OECD study of more than 5,000 small and medium-sized businesses found that 31% were already using generative AI. Among users, 29% said it helped them compete with larger companies.
Okay, let’s up the ante. Imagine a large corporation. I know, evil, right? But this particular company produces sustainable products that genuinely improve people’s lives and reduce environmental harm. It is also experiencing serious financial problems.
Management determines that the company has two realistic options. It can continue operating as it has and potentially go under, eliminating hundreds of jobs and the products it creates, or it can automate portions of its design and marketing operations. Some designers will lose their jobs, but the company and most of its employees will survive.
What about this case? If the company continues as is, the likely result is not only greater job loss but the disappearance of products that provide some broader social benefit. In this scenario, are we still prepared to say that using AI to replace design labor is inherently wrong?
These examples are not meant to dismiss designers who have strong negative opinions about AI. These thought experiments are meant to expose a problem with how we talk about the ethics of AI and creative work. Not every situation in which AI is used to create a design is malicious, exploitative, or even particularly interesting ethically.
More importantly, the examples reveal that most of us probably do not believe that replacing human design labor with AI is inherently unethical. If we did, our judgment should remain relatively consistent across all these scenarios. Yet I suspect it does not for many people. As the circumstances change, so does our judgment, even though the underlying technological act remains largely the same.
This becomes clearer if we change the corporate example again. Imagine the same company is not struggling at all. It is enormously profitable and eliminates its design staff because AI can perform enough of their work at a lower cost. The savings go directly toward increasing profits. I suspect many designers who tolerated the previous examples would object much more strongly here.
But that reaction tells us something important. The technology has not changed. The displacement of human labor has not changed, at least compared with the previous corporate example. What changed were the circumstances, the apparent necessity of the decision, and who ultimately benefited. If replacing human creative labor with AI were itself the ethical violation, these differences should matter far less than they apparently do.
To make things interesting, let’s complicate the scenario. Suppose the profitable corporation uses the savings from automation to lower prices, increase wages for its remaining employees, or expand access to products that improve millions of people’s lives. The designers who lose their jobs have not suddenly become irrelevant, but they are no longer the only people with something at stake.
We are now weighing their interests against those of employees, customers, communities, owners, and perhaps society more broadly. What initially looked like a simple question about whether AI should replace designers has become a question about how benefits and harms should be distributed when technology changes the economics of human labor.
This is why I think asking whether AI-generated design is ethical is probably the wrong question. The more difficult question is under what conditions its use becomes unjust. There are really two questions here. First, was the decision justified? Was automation necessary, were reasonable alternatives available, and what resources did the person or organization have?
Second, how were the consequences distributed? Who benefited from the increased efficiency, who absorbed the harm, and what obligations remained toward workers whose labor could now be automated?
A person creating a missing-dog poster and a multinational corporation eliminating an entire design department may use the same technology for essentially the same functional purpose, but that does not mean the two acts are morally equivalent.
These examples involve two different things, and they should not be confused. In some cases, AI is replacing an existing designer. Someone loses a job or a contract that otherwise would have existed. That raises relevant questions about labor, responsibility, and how to distribute automation’s benefits.
But in other cases, no designer has actually been displaced. The person making a missing-dog poster probably never would have hired a designer. The struggling pizza shop may not have been able to afford one. AI did not take that work from a designer because there may never have been paid design work available in the first place. What changed is that someone who lacked access to professional design gained the ability to create something functional for themselves.
This is where designers’ response becomes particularly interesting. Design is a profession that talks endlessly about empathy. We are taught to understand people’s circumstances, recognize constraints, reduce barriers, and design around the realities of people’s lives. Yet when someone without the money or skills to hire us uses AI to overcome precisely those constraints, some designers suddenly become much less interested in empathy.
There is an uncomfortable contradiction there. If we believe design should empower people, we have to at least confront the possibility that AI is doing exactly that for some people. We cannot celebrate accessibility and democratization when designers provide them, then condemn those same principles when a technology reduces someone’s dependence on designers.
With all this talk about AI replacing design jobs, a designer might object that I am ignoring something fundamental. AI is not simply another technology that produces design work. It was trained on enormous amounts of existing human creative work, often without the knowledge or consent of the people who created it.
So the problem may not be that AI replaces designers. Perhaps the technology itself depends on designers whose work made it possible in the first place. The U.S. Copyright Office’s report on generative AI training shows how questions about copying, fair use, and licensing have become part of the broader debate.
That is a legitimate objection, but it opens another philosophical problem. Designers do not create in a vacuum either. We learn by studying existing work, absorb visual conventions, imitate styles, respond to trends, and recombine ideas that came before us. Roland Barthes pushed this problem much further in The Death of the Author, challenging the idea of the creator as the singular origin of a work. Although Barthes wrote about literature, his argument has been applied more broadly to questions of authorship and creative production. Generative AI makes that old problem newly complicated. Where exactly does influence end, authorship begin, and appropriation become exploitation?
None of this means that human learning and AI training are equivalent. The scale, method, consent, compensation, and commercial use involved raise legitimate ethical questions. But following that argument takes us into a much larger debate about originality, ownership, and whether creative work can ever be completely separated from what preceded it. That is an important debate, but it is not the one I am exploring here. My concern is narrower. Given that AI-generated design exists and is being used, under what circumstances is its use actually unjust?
The economic effect of AI reducing the need for specialized outside labor is already visible in small businesses. OECD research found that 39% of generative-AI-using SMEs that had experienced a skill gap said the technology helped compensate for it. The same report found that 14% of SMEs using generative AI had reduced their reliance on outside contractors.
Treating every use of AI-generated design as exploitation ignores the capabilities the technology provides to people with limited resources. Treating every use as technological progress ignores the very real consequences for people whose livelihoods depend on work that can now be automated. Neither position gets us very far.
No clean line separates ethical from unethical uses of AI in design. Designers have a legitimate interest in protecting the value of their labor, but that interest cannot establish an obligation for everyone else to purchase that labor. Carriage makers also had a legitimate interest in protecting their livelihoods, but that did not create an obligation for society to keep buying carriages after the automobile changed transportation. The uncomfortable reality is that labor value does not guarantee permanent demand. Businesses have legitimate reasons to pursue efficiency, but efficiency cannot justify every decision simply because it reduces costs.
AI is making that conflict difficult to ignore because it is dramatically changing who can produce creative work and what that production costs. But the underlying ethical problem is larger than AI and larger than design. Technologies have always redistributed human capabilities, labor, and economic value.
The question, then, is not whether machines should ever do work that humans once did. It is whether replacing that labor is justified by the circumstances, who benefits from the decision, and who bears the cost. The lost-dog poster is easy. The profitable corporation is harder. But that is precisely the point. If our judgment about AI changes as the circumstances change, then the ethical question was never simply whether AI replaced a designer. It was why, for whose benefit, and at whose expense.
Don’t miss out! Join my email list and receive the latest content.
Lost dog—or lost designer? was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.
