The most honest thing anyone can say about artificial intelligence and creative work in 2025 is also the least satisfying: it depends. It depends on what kind of creative work you do, what value you bring to it, how your clients understand and compensate that value, and how quickly the technology continues to develop in directions nobody can fully predict. If you were hoping for a confident narrative — “AI will free creatives to do their best work!” or “AI will replace most creative jobs within five years!” — this piece will disappoint you. Both of those narratives are being sold by people with financial interests in one conclusion or the other, and both are probably wrong in the ways that matter most.
What is not in doubt is that AI has already changed creative work, is changing it further as you read this, and will continue to change it in ways that require the creative industry to think carefully — more carefully than it has so far — about what it is actually selling. Because the answer to the collaboration-versus-displacement question depends entirely on what you believe the value of creative work actually consists of. And that’s a conversation the industry has been avoiding for decades.
What AI Is Actually Good At (So Far)
Current AI systems — language models, image generators, video tools — are exceptionally good at producing competent, fluent, contextually appropriate creative output at high speed and very low cost. This is not a small capability. “Competent, fluent, and contextually appropriate” describes a significant portion of what the creative industry produces every day: the social media caption, the product description, the email newsletter, the banner adaptation, the stock-photography-style brand image, the first draft of the press release.
This work has always existed at the lower-value end of the creative spectrum — it was competent rather than distinctive, efficient rather than resonant, correct rather than surprising. AI performs it well precisely because “correct” and “contextually appropriate” are things that can be learned from enormous amounts of training data. The pattern recognition that underlies most language and image generation is a genuine form of intelligence. It is not, so far, the form of intelligence that produces work that surprises you, that makes you see something you hadn’t seen before, that changes how you think about a problem. But it is very useful for everything below that threshold.
What AI Is Actually Bad At (So Far)
AI systems are poor at genuine novelty — at producing ideas that aren’t recombinations of patterns in their training data. This is sometimes masked by the fact that humans also produce mostly recombinations of existing patterns, but humans can draw on lived experience, embodied knowledge, cultural context, and emotional intelligence in ways that current AI systems do not. The experienced creative director who has worked in a specific category for fifteen years brings knowledge that can’t be extracted from text and image corpora. That knowledge is worth something. The question is whether it’s worth enough.
AI is also poor at the relational and judgment dimensions of creative work — at understanding what a specific client in a specific organizational context actually needs, at navigating the human dynamics of a creative process, at knowing which battle to fight and which to let go, at building the trust that allows genuinely risky creative work to be commissioned and approved. As we’ve explored throughout this journal, from the stakeholder syndrome to the art of saying no, creative work is not only a technical act. It’s a relational and political one. And that dimension is not something current AI systems can replicate.
The Value Conversation the Industry Isn’t Having
The arrival of AI has exposed something the creative industry should have been examining more honestly for years: a significant portion of what it charges for is commodity, not craft. The brief adaptation, the format resize, the copy variation, the stock-library-adjacent visual — these have been priced and positioned as creative services when they are, in many cases, administrative tasks that require creative skill but not creative judgment. AI can do them. And clients know it, even when agencies and freelancers haven’t yet fully updated their pricing conversations accordingly.
The creatives who are navigating AI most successfully are the ones who have been honest with themselves — and their clients — about which parts of their work are commodity and which parts are genuinely differentiated. The commodity parts can be augmented with AI to increase efficiency and reduce cost. The differentiated parts — the strategic insight, the original concept, the cultural knowledge, the relational skill, the judgment that comes from experience — remain human, for now, and remain valuable precisely because AI’s ability to replicate them is limited.
As we’ve argued in the context of what makes creative work actually effective, the value of creative work has never been primarily in its production — it’s been in the thinking that precedes production. AI can accelerate the production. It cannot yet replicate the thinking. Which means the most important thing any creative professional can do right now is invest in the thinking: in the strategic, cultural, and relational skills that make production worth doing.
The Displacement That’s Already Happening
Let’s be honest: displacement is happening. Not in the dramatic way the most alarming headlines suggested — mass creative unemployment, agencies collapsing, the end of the profession — but in the quieter, more structural way that matters more over time. Entry-level creative work is being reduced in some organizations because tasks that used to require a junior are now being done by a senior with an AI tool. The hiring pipeline at the bottom of the creative industry is narrowing in ways that will affect the development of the next generation of senior talent — because seniors come from juniors, and junior creative work is where foundational skills are built.
This is a real problem and it doesn’t have an easy solution. The industry that fails to invest in developing new creative talent because AI can handle the entry-level work will find itself, in ten years, without the experienced senior talent it needs — because that talent wasn’t developed at the junior level when it should have been. It’s the same logic as the burnout problem: systemic short-term decisions with long-term costs that aren’t visible until they’re very expensive.
The Collaboration That’s Actually Worth Building
The most useful version of the AI-creativity relationship isn’t the one where AI generates and humans approve. It’s the one where humans do the thinking and AI handles the execution of decisions that have already been made through genuinely human judgment. The creative director who uses AI to rapidly visualize five different directions from a strategy they’ve already worked out is using the tool well. The creative director who uses AI to generate the directions and then picks the one they like is outsourcing the part of their job that was worth doing.
The distinction matters because what we’re building, in how we use these tools, is a set of habits and capabilities that will define what creative professionals are and are not able to do in the next decade. Build the habit of using AI to execute human thinking, and you end up with faster, more capable human thinkers. Build the habit of using AI to replace human thinking, and you end up with a generation of curators rather than creators. Both can survive in the short term. Only one produces work worth caring about in the long term.
Still figuring out where AI fits in your creative practice? Our shop is for people who care about making things worth making — with or without the machines. Probably both. Definitely on your own terms.
