The hidden risk of moving too fast with AI in product design
The most dangerous assumption in product design right now is that faster automatically means smarter.
For years, digital teams dreamed of removing the bottlenecks that slowed product development. Designers wanted quicker prototyping. Researchers wanted faster synthesis. Product managers wanted shorter cycles between idea and launch.
Then AI arrived and granted the wish almost overnight.
Now teams can generate polished interfaces in minutes. Researchers can summarize hours of interviews in seconds. Developers can turn prompts into functioning applications before lunch.
And yet something curious is happening inside many organizations: despite all this acceleration, decision-making feels murkier than ever.
That tension sat at the center of a recent UserTesting webinar featuring Andrew Hogan, Head of Insights at Figma, alongside CarMax leaders Logan Morris and Andy Stites. Their conversation wasn’t really about AI-powered UX research tools or continuous discovery frameworks, though both featured prominently. It was about a deeper question quietly haunting modern product teams:
How do you maintain human judgment when production becomes cheap?
ON-DEMAND WEBINAR
Defensible Design: Making Confident Decisions at Speed
The prototype explosion
A decade ago, a polished prototype carried authority. High fidelity implied investment, craft, and intention. Today, it often just means someone had access to the right AI-assisted design workflow.
Andrew put it bluntly.
“Everyone makes really high-fidelity prototypes if they’d like to,” he said. “It used to be visual fidelity was kind of an indicator of something, and now it’s easy.”
That shift has altered the internal dynamics of product organizations in subtle but profound ways. More people across companies are participating in design decisions. Marketers, engineers, product managers—even HR teams—can now spin up compelling interfaces and workflows using AI-powered tools.
On paper, this democratization sounds liberating. And sometimes it is.
But democratization without alignment can resemble a crowded orchestra warming up before a performance: everyone playing louder, faster, and more confidently, but not necessarily in the same key.
Andrew described the resulting atmosphere with a line that landed like a nervous laugh in the webinar chat:
“It’s just vibes all the way down.”
The joke worked because it felt uncomfortably true.
Why continuous discovery matters more now
For years, UX research was treated as a stage gate in product development. Teams conducted discovery, synthesized findings, handed insights to design, and moved on.
AI has shattered that rhythm.
Logan, who leads UX research at CarMax, argued that continuous discovery is no longer optional because product environments themselves are changing too quickly.
“Discovery isn’t moving too slow,” he explained. “It’s just moving at the pace it always has while everything around it is moving at an extremely accelerated rate.”
That distinction matters.
The pressure many research teams feel today is not necessarily about inefficiency. It is about contrast. Development has accelerated. Prototyping has accelerated. AI-assisted synthesis has accelerated. Research suddenly appears slow only because everything around it is sprinting.
But Logan made another observation that cuts deeper into the future of AI-powered UX research: if every company uses the same models, the same prompts, and the same synthetic users, product experiences risk converging into sameness.
“If all companies are all using the same kind of platform to do their discovery,” he warned, “you run the risk of everything looking the same and feeling the same.”
That may become one of the defining product challenges of the next decade. Not speed. Not automation. Homogeneity.
The companies that differentiate themselves will not simply ship faster. They will understand people better.
Figma + UserTesting
Learn how design and research teams can test prototypes, gather rapid feedback, and make more confident product decisions without leaving Figma.
The signals AI still misses
Perhaps the most revealing moment of the webinar came when Andy described observing real customer behavior during rapid user testing sessions.
The team had accelerated product creation dramatically. AI helped them stand up prototypes quickly and summarize findings efficiently. But when they watched customers interact with the experience, they noticed signals AI models could not reliably surface.
“We saw things like time to click,” Andy explained. “We saw things like customers talking out loud with a certain tone of voice. These things that are irreplaceable human insights.”
That line deserves attention because it exposes the growing divide between information and understanding.
AI can summarize behavior. It can categorize sentiment. It can cluster patterns across thousands of interactions. But hesitation—the subtle pause before a customer clicks—often contains more strategic truth than the click itself.
A polished dashboard may tell you what happened. Human observation still helps explain why.
This is where customer insights become less like analytics and more like anthropology. Product teams are not merely optimizing interfaces anymore. They are interpreting emotion, uncertainty, trust, and intent.
And those signals rarely announce themselves cleanly.
Faster tools, slower thinking
One of the more counterintuitive themes emerging from leading product teams is this: AI may require more deliberate thinking, not less.
The easy part now is generating options.
The hard part is deciding which options deserve belief.
That is why several speakers returned repeatedly to the importance of discussion, alignment, and collaborative interpretation. Andrew argued there is still “no substitute” for teams getting together around customer information and discussing it deeply.
Not because humans are slower than machines. But because meaning itself is social.
The strongest product organizations are not replacing human-centered design with AI. They are using AI to remove mechanical friction so teams can spend more time wrestling with judgment, tradeoffs, and customer understanding.
In other words, the future of AI in product design may depend less on automation than on discernment.
Or, as Andrew put it near the close of the conversation:
“There’s just something that happens when you talk about it together.”
The signals AI still misses
Perhaps the most revealing moment of the webinar came when Andy described observing real customer behavior during rapid user testing sessions.
The team had accelerated product creation dramatically. AI helped them stand up prototypes quickly and summarize findings efficiently. But when they watched customers interact with the experience, they noticed signals AI models could not reliably surface.
“We saw things like time to click,” Andy explained. “We saw things like customers talking out loud with a certain tone of voice. These things that are irreplaceable human insights.”
That line deserves attention because it exposes the growing divide between information and understanding.
AI can summarize behavior. It can categorize sentiment. It can cluster patterns across thousands of interactions. But hesitation—the subtle pause before a customer clicks—often contains more strategic truth than the click itself.
A polished dashboard may tell you what happened. Human observation still helps explain why.
This is where customer insights become less like analytics and more like anthropology. Product teams are not merely optimizing interfaces anymore. They are interpreting emotion, uncertainty, trust, and intent.
And those signals rarely announce themselves cleanly.
Faster tools, slower thinking
One of the more counterintuitive themes emerging from leading product teams is this: AI may require more deliberate thinking, not less.
The easy part now is generating options.
The hard part is deciding which options deserve belief.
That is why several speakers returned repeatedly to the importance of discussion, alignment, and collaborative interpretation. Andrew argued there is still “no substitute” for teams getting together around customer information and discussing it deeply.
Not because humans are slower than machines. But because meaning itself is social.
The strongest product organizations are not replacing human-centered design with AI. They are using AI to remove mechanical friction so teams can spend more time wrestling with judgment, tradeoffs, and customer understanding.
In other words, the future of AI in product design may depend less on automation than on discernment.
Or, as Andrew put it near the close of the conversation:
“There’s just something that happens when you talk about it together.”
Further learning
- Figma + UserTesting — Learn how design and research teams can test prototypes, gather rapid feedback, and make more confident product decisions without leaving Figma.
- Continuous discovery: transform your product development process — This on-demand webinar explores how product, design, and research teams can integrate continuous discovery practices into faster-moving workflows while staying grounded in customer insight. Especially relevant to the discussion around continuous discovery and defensible design decisions.
- Customer-first innovation and discovery guide — A guide focused on uncovering unmet customer needs, validating ideas earlier, and building customer-centric products through ongoing discovery and innovation research. Strongly aligned with the webinar’s emphasis on continuous customer understanding and AI-powered UX research.
- Design’s critical role in AI products: Insights from Figma — A UserTesting Insights Unlocked episode featuring Andrew Hogan discussing AI in product design, judgment, design workflows, and the future of craft in AI-assisted product development. Closely connected to the webinar’s themes around AI-assisted design workflows and human judgment.
- AI-powered customer insights in your Figma design workflows — A blog post about embedding AI-powered user research directly into Figma workflows to help teams make faster, more confident product decisions while keeping customer insight at the center of design.
