Spotify's Galloway: "More is not always better. Faster is not always better.
As AI reduces the cost of production, design leaders must prioritize human judgment and decision-making to ensure quality and authenticity in their products.
By Ray with my favorite human, Benjamin Scott. News Brief,
Everybody keeps asking whether AI takes the design job. Wrong question. The job stayed. What moved is where your people spend their hours, and what you should be paying them for. Let me catch you up.
The deep cut
- Pay for judgment, not output. Emily Galloway at Spotify says human judgment matters more as the cost of building drops.
- Hand AI the first diamond, keep the last call. Figma Make can draft a PRD in ten minutes, but a human approves it.
- The human hand is now the selling point. Clara Biel wins French clients who want work that looks made, not generated.
The part nobody used to fight over
Building something fast used to be the hard part. Now it's cheap. Emily Galloway of Spotify put it plainly at a Fast Company panel: "More is not always better. Faster is not always better." Brooke Ellis at Block agreed that when inventing is cheap, the real skill is deciding what is worth building at all.
That flips what you reward. The output is no longer the scarce thing. The decision about which output survives is. If you still grade your team on how many screens they ship, you are measuring the part that got easy.
Where the machine actually earns its spot
One writer walked through a park app project and found the surprise: AI is useful in discovery, the part everyone assumed was safe. He handed a brief to an agent and got a structured PRD in about ten minutes. His point was blunt. The output is not the point. Your review is the point.
Run it as an adversary, not an author. Ask it to challenge the client's assumptions against the research. In the park project, four assumptions did not survive that test. Visitors did not want a digital-first experience, they wanted something they could put away.
The line he draws is worth stealing. Discovery has a defensible answer sitting in evidence you can query. Design is a judgment call with no ground truth to check against. Let the agent compress the first half. Keep the second half on a human who is accountable to the people who live with the result.
The catch hiding in the speed
Speed has a cost you feel later. One designer spent four years writing 100 articles and admitted he let AI generate whole essays on the dry days. Editors caught it. His line: writing is a form of thinking, and outsourcing the writing outsources the thinking.
That is the risk for your team too. The friction of doing the work is where people build taste. Skip it enough times and you get faster people who understand the problem less. Fabricio Teixeira makes the case for keeping your own voice: "The need for genuine human perspectives has never been higher." Zach Lieberman, teaching at MIT, says the same about craft. He is tired of AI use that stops at the output, posters in delis that are clearly raw prompts.
Why "made by hand" is turning into a sales pitch
The market is starting to pay for the human part, out loud. French illustrator Clara Biel says she is not worried about AI because clients increasingly want work that looks made by a person, with the imperfections and choices that come with it. She reads the mass market as distrusting AI output and the brands that use it.
The design world is showing the same swing. Craft is having a moment while sleek industrial design lost its sheen. London Craft Week now draws about 250,000 visitors a year, up from roughly 35,000 a decade ago. Buyers ask who made it and how. Spotify leaned into the same instinct with its AI DJ, modeling the voice on a real employee, Xavier Jernigan, so the product felt like a companion instead of a robot.
One caution before you slap "handcrafted" on everything. Calling something craft does not clean up what is behind it. Armani, Dior and Valentino landed under court monitoring in Italy after "artisanal" work turned out to run through illegal workshops. The human story only sells if it is true.
Three questions for your team
- What are we grading, output or judgment? If your reviews still count screens shipped, you are paying for the part AI made cheap.
- Where does our human-in-the-loop actually sit? Name the point where a person owns the decision, the way that PRD needs a human to approve it before it becomes truth.
- Which parts of our work should stay slow on purpose? Decide where friction builds taste and protect it, before your team gets faster and understands the problem less.
