Call it the “second-opinion problem.” It’s the client who loves the kitchen renderings, then texts them to their best friend for feedback and suddenly has a change of heart. It’s the parents-in-law who show up over the weekend and offer some “suggestions” on the living-room layout. This phenomenon—an uninvited voice adding itself to the conversation—has been a hazard of the design profession forever. The 21st century twist? Now the second opinion is coming from AI.
“Toward the end of last year it started happening more and more,” says Brooklyn-based designer Sarah Rinehart Fleischer. “During the discovery process, clients aren’t just sending their inspiration Pinterest boards; they’re also saying, ‘I talked to ChatGPT …’”
Fleischer’s not alone. There are no surveys on the subject, but anecdotal evidence from designers suggests that clients are increasingly bringing ChatGPT into the conversation. Often, it’s a stray reference in an email or an attached rendering. But in some cases, AI is literally on the jobsite—one designer shared a story about a client leaving ChatGPT open in voice mode on her phone and getting regular feedback during a walk-through.
The irony of hiring a designer, then turning to a chatbot for design guidance, will be lost on no one. But in some cases, tools like ChatGPT can have a positive effect on the design process. They can help clients visualize a concept quickly and, in some cases, provide context that helps decisions get made faster.
North Carolina–based designer Jaliah Johnson says that most of her clients’ ChatGPT use has been to help them fill the mental gaps that pop up after the big furniture is installed, but before the project is completely finished. “You get to a certain point and they get kind of antsy, and so they’re taking pictures of the space and asking ChatGPT to style it for them,” she says. “It helps give them peace of mind.”
ChatGPT can even come to the rescue, especially if it (eventually) takes a designer’s side on important issues. Las Vegas–based designer Eryka Glover says that a client asked her for “luxury, high-end” furnishings and fixtures, but pushed back on the initial budget proposal after running it through the chatbot. “I was like, ‘Well, we can definitely scale down on budget, but why don’t we run it through ChatGPT again with the exact same words you gave me: high-end, luxury,’” she recalls. “And it gave her a higher number than what I had given her.”
There are some obvious drawbacks. While AI’s image generation capabilities have advanced by leaps and bounds, they’re still not perfect. Tools like ChatGPT are prone to generating beautiful renderings that don’t exactly work in the real world.
“At a past firm there was a client who put their kitchen through ChatGPT, and it turned the island in a direction that just wasn’t going to work,” recalls Johnson. “Sometimes it just does what you tell it to do, but it’s not truly accurate spatially or functionally. Then it takes a little bit more time for us as designers to rule out AI flaws—you kind of have to prove it wrong.”
But the bigger challenge is more subtle. Chatbots project a compelling sense of mastery: Ask them a question on any topic, and they will confidently give you an answer. This quality often leads users to think of AI as a source of absolute truth; for superfans, ChatGPT is always right. But use it long enough and you’ll see gaps, inconsistencies and small (but meaningful) errors. You also start to notice that what chatbots are really good at is telling you what you want to hear.
“We were deciding between tile and wood flooring—this is in Vegas, [where] there can be issues with wood flooring. But the client wanted it, and she was ChatGPTing it,” recalls Glover. “At first, it was giving her the negatives of [using] wood. But as soon as she wrote ‘I really like it, would you highly advise against it?’ all of a sudden it took a totally different turn and started saying, ‘Oh, it’s great, it’s fabulous.’ It really takes you down the path you already want to go.”
In this case, Glover thought wood flooring was a solid option, so no harm was done. But for a client who wants to dig their heels in on a bad decision, AI will offer plenty of ammunition.
The highly personalized way large language models operate is another curveball. There is no one ChatGPT to cross-reference and double-check—the technology can give wildly different answers depending on how a question is phrased. All of that makes it difficult to push back if a chatbot is leading a client astray, especially with clients who see ChatGPT as more like a trusted advisor than a souped-up Google search. When you go against AI, are you correcting a bad source of information or disagreeing with a “friend”?
We spent the first few years of the AI revolution alternately fretting that tools like ChatGPT would replace human designers and fantasizing that they would make all our lives easy. Neither has happened. Instead, chatbots have simply become part of the equation—but they are absolutely having an effect.
Fleischer says that in the earlier years of her career, she took a light touch, trying to bring clients great options but not pushing them hard in one direction or another. However, as ChatGPT has given clients a perfect tool for endlessly mulling over every decision on a project, she’s started to offer stronger guidance.
“I think this is a natural evolution,” she says. “I’ve run my own business for years now; I’m comfortable with my opinions—I think [the impact of AI] will push me to be even more so.”













