Opinion
The Machine Can Draw. It Still Can't Design.
Generative AI is genuinely useful and genuinely flawed, and the reason it is not taking your design job yet comes down to the difference between making an image and solving a problem.
Type a sentence, wait a few seconds, and a striking image appears. A few years ago that felt like magic. Now it is Tuesday. For a graphic designer, the reaction is understandable and very human: if a machine can conjure that from a prompt, what exactly is left for me? The honest answer is that a great deal is left, though not always for the comforting reasons people reach for. Generative AI is more useful than its critics admit and more limited than its boosters pretend, and the gap between those two things is where your job still lives.
Start with what it does well, because pretending otherwise fools no one. Generative image tools are a genuine cure for the blank page. They spin up moodboards in minutes, throw twenty directions at a concept before lunch, and produce backgrounds, textures, and placeholder assets that used to eat an afternoon. They let a one-person shop show a client three visual worlds instead of describing one. For small studios and non-designers especially, they put a kind of visual fluency within reach that used to require a whole team. That is real, and it is not going away.
But look closely and the limits arrive quickly. The first is sameness. These models are trained on the average of everything, and they pull, relentlessly, toward a glossy mean. Ask ten people for a modern minimalist logo and the machine will hand all ten something that rhymes. Distinctiveness is the entire point of design, and the tool is quietly working against it. A look that everyone can generate is a look that is worth less the moment everyone does.
The second limit is one most people never see until it matters: ownership. Under current US law, an image generated purely from text prompts is not eligible for copyright protection. The Copyright Office reaffirmed exactly this in its January 2025 report, concluding that prompts alone do not give a person enough control over the result to count as authorship, and the courts have upheld the human-authorship requirement. Translated into shop terms, if you deliver pure AI output to a client, neither of you may own anything protectable in it. For a logo, a campaign, or a package, that is not a footnote. It is a business risk with the client's brand attached to it.
The third limit is the one every working designer already feels. AI gets you to eighty percent fast and then fights you for the rest. It cannot reliably set legible, correct type. It drifts off brand color. It reshapes a logo that is supposed to stay fixed. It invents details, garbles small text, and has no idea whether the result holds up at business-card size and billboard size both. The closer you get to finished, accountable, on-brand work, the more human hands it takes, and the part it cannot do is precisely the part clients pay for.
Which points at the real reason your job is safe for now. Generation is not design. A prompt is a wish; a brief is a problem. Design is deciding what the thing is for, who it speaks to, what it must not do, how it fits a brand, and whether it is even the right solution, and then being accountable for that judgment to a person trusting you with their business. The machine can fill a canvas. It cannot decide what the canvas is for, judge whether the answer is right, or stand behind it when the client asks why. Clients are not buying pixels. They are buying judgment and trust, and those do not come out of a prompt box.
None of this means nothing changes. It means the work shifts. The commodity tasks compress: the quick stock-style image, the fast variation, the production hours a junior used to bill all get cheaper and faster. Value migrates upward, toward taste, direction, strategy, and the skill of directing the tool well. The threat to designers was never a machine that draws. It is the designer down the street who learned to use that machine as an instrument while you argued about whether it counts as art, and the client who has been convinced that the pixels were the product. The right response is not fear. It is to become the director rather than the draftsman.
So let the machine draw. It is good at drawing, and getting better. It still cannot design, because design was never really about drawing. It was about deciding, deciding on someone else's behalf, and answering for the decision. Use the tool to reach the decisions faster, and keep your hands on the part that was always the job. For now, and for as far ahead as now reaches, that part is yours.
About this piece. This is an opinion piece on creative work and technology; it is not legal advice. The copyright treatment of AI-assisted work is evolving and fact-specific, and anyone with a real ownership question should consult qualified counsel.