Fluency and Experience: How UX Researchers Across Career Stages Are Experiencing AI

Welcome back to Think First: Perspectives on Research with AI. There’s a real tension in UX research right now. On one hand, junior researchers are sometimes running circles around veterans when it comes to AI adoption. On the other, there’s a growing perception that entry-level research jobs are disappearing entirely. So which is it – is AI opening a door for people starting out, or closing one?

We put that question to three researchers at different career stages. Peter Aldous, a Senior UX Researcher at Microsoft, is the balanced veteran who believes that AI is a tool, and the more you know, the better you can wield it. Isis Chen, a UX Researcher on the Xbox Research team, is earlier in career, describing AI as the 24/7 mentor she never had. And Hilary Dwyer, a Principal UX Researcher, is the senior strategist reframing the whole role around judgment and foresight.

The pattern that emerges is subtler than youth vs experience. AI fluency is fast to acquire; experience is slow. What AI accelerates is the gathering of knowledge – what it can’t shortcut is the judgment that only years of building real products with real people produce. On that, the junior and the two seniors agree.

Opportunity, or the door closing?

We opened each conversation with this question.

Peter’s answer was that both things are true at once. “There are fewer entry-level jobs,” he says – the market is tight industry-wide, and budgets are thin – “but you also see the folks who tend to be a little bit newer in the field… diving into AI the most, and they are kind of running circles around other folks.” The two facts seem opposed, he notes, but we’re simply seeing a little of both.

Isis, is more openly optimistic – because the technology has changed how fast she can grow. Her team gives her plenty of more experienced mentors, “but of course everyone’s very, very busy… nobody’s going to be available 24/7.” AI, grounded in the team’s own past research and methods, becomes “a mentor that’s accessible 24/7,” one she can ask “my so-called dumb question” without worrying she’ll “leave a bad impression” on a more senior teammate. She’s clear-eyed about the counter-view – that to leadership, “if the senior can do a lot more work than junior researchers plus the AI,” juniors can look less necessary – but pushes back: “every generation you need researchers,” and this is a moment to grow them faster.

Hilary calls the framing “more nuanced.” What used to be a regular path in for entry-level researchers “is different now”; being AI-first and being able to “stay curious longer” matter more. She won’t say the door is closing – but she does think there’s “a real consensus that entry-level research jobs are disappearing,” and is careful about the cause: “it’s not because of AI. It’s because there’s less money for those head counts now than there used to be.” And even so, she adds, “we also really, really need smart, intuitive, compassionate people asking tough questions… researching how on earth AI is changing everything that we do.”

The 24/7 mentor – and what it can’t teach

Isis is the clearest on what AI gives someone who is starting out. It’s “like an encyclopedia, like Wikipedia, but much, much better” – and, trained on her team’s materials, it teaches domain expertise at a pace no busy mentor could match. But she’s just as clear on what it doesn’t build: “critical evaluation” – knowing “what is good, what is not so good, why” – which “really comes from experience.” AI “would produce something that’s very convincing… sounds really professional,” and without enough experience it’s hard for a junior to challenge it: “That confidence comes from years of expertise.”

Peter agrees the newer researchers are adopting fastest, but warns the leverage runs through knowledge. On quant work, prompt the model precisely and “you get a much better result”; throw in data and say “compare these two groups” and “it will do it, but it might not be the right test.” If you don’t know “the difference between parametric and non-parametric stats and when you need to use them,” you won’t even know how to answer the questions the AI asks you. His blunt version: “A crap question gets you a crap response, period… with AI it just gets you that bad response more quickly” – and AI “tends to make that bad response look better and more believable.”

Hilary sees a related failure mode, and it isn’t limited to those with less experience. AI has “shifted our language to be fancier, more ivory tower” – it returns “long words like overreliance or strategy or human-centeredness,” which “we start to toss around” with everyone meaning different things. Material clearly written by AI, she notes, often “doesn’t survive when you critique it or ask for more specifics.” That happens “across all researchers, no matter the level” – though younger researchers, more inclined to reach for AI on the things that make them nervous, may show it more.

Is there a ceiling?

If you learn the craft with AI from day one, does your growth eventually cap out? All three said no.

“I don’t see it as a limitation – I see AI as a tool,” Peter says, so he doesn’t see a ceiling “as long as you’re doing the learning on the other side as well.” The danger is dependence: “If you come to use it solely as a crutch, then it will be limiting… but if you use it to expand your capabilities” while still learning the methods underneath. AI, he adds, “has to be vetted and verified,” and you can’t vet what you don’t understand.

Isis frames the ceiling as a knowledge horizon: “We don’t know what we don’t know until we learn about it.” AI “can tell you everything that you ask – but how do you know what questions to ask? How do you know the scope that’s outside of your current knowledge?” Developing a point of view on a field “takes a lot of digging, a lot of in-depth learning”; AI “speeds up learning, but it does not replace the growth.”

Hilary captures it in a word: curiosity. There’s no ceiling “as long as you are pushing to stay really curious” – “that’s what always has made researchers great,” whether your background is academic, applied, or brand-new with AI’s help. Keep asking good questions and “keep going to other people who might know more than you, because it is very easy to get off the rails.”

Tedious work, or work you should still know how to do?

Everyone wants AI to absorb the drudgery. Drawing the line turned out to be personal.

Peter, finds AI still isn’t great at the truly tedious parts – it summarizes and organizes well but stumbles at formatting a deck even from a template, leaving him to “reconnect and format everything” by hand. Where he draws a firm line is the researcher’s craft. “One of the researcher’s superpowers is being able to make connections between things and read between the lines. AI can’t do that.” Nor can it observe: in a usability study, a participant’s “body language is saying this is utter crap and I hate this product,” while their words say “it can be better, but this is actually pretty good.” AI takes the words at face value; a researcher reads the room and probes.

“I’m a researcher at heart,” Isis says, so she draws her line by asking what she actually loves. Early on she assumed AI would teach her to code; instead “it just codes for me,” and she realized she enjoys “the capability of coding” but not the process. So she lets AI own the process she doesn’t care about, once she can trust the output, and reinvests the time in the part she does care about: insights. The result, she says, is counterintuitive – “a lot higher cognitive load,” with far less of the “mental downtime” she used to spend “fiddling with the Excel sheet.” It’s why she believes AI can make researchers think deeper, not shallower.

“I can’t use AI unless I have decided what my point of view is on something,” [Hilary says – a rule that inverts the usual order. So she reads the transcripts herself, writes a little herself, and only then brings AI in to flesh it out. “Any time that I’ve gone to AI first, it leads to things where I can’t think as big – and it doesn’t even feel like I’m getting the tedious stuff done.”

What AI won’t close

Asked to define the gap in experience that AI will never close, all three pointed at the same thing: accumulated knowledge, and the judgment it produces. For Peter, it’s “those higher-level skill sets – the ability to make connections between things… to know what’s important about one type of method, or where to use it,” and to “pull all of that together into a great plan that’s going to get you good answers that are valid.” For Isis, it’s “years of experience” that stop being “a list of things” and become “second nature” – an intuition she can’t always articulate but that lets a researcher “ask the right question” and “know what it doesn’t know.” For Hilary, it’s the experience of “building products and finding ways to be useful” to designers, PMs, and engineers – “a lot of empathy that grows over time,” and “knowing exactly what value means to the people you’re supporting.” She casts the senior’s edge as a new kind of contribution, too: “principled judgment” – having a point of view at the table about what people really need – and “strategic foresight,” thinking twelve to eighteen months out so insights are durable rather than one-offs.

Coming up next

Starting out or veteran, the three keep circling the same conclusion: AI closes the knowledge gap quickly and the experience gap not at all. It can be a 24/7 mentor, an accelerant, a tireless first draft – but the judgment to ask the right question, read the room, and know what actually matters is still earned the slow way. As Hilary put it, that’s why “we need researchers now more than ever.” In the next episode, we turn to another corner of the practice being remade around AI.

Welcome back to Think First: Perspectives on Research with AI. There’s a real tension in UX research right now. On one hand, junior researchers are sometimes running circles around veterans when it comes to AI adoption. On the other, there’s a growing perception that entry-level research jobs are disappearing entirely. So which is it – is AI opening a door for people starting out, or closing one?

We put that question to three researchers at different career stages. Peter Aldous, a Senior UX Researcher at Microsoft, is the balanced veteran who believes that AI is a tool, and the more you know, the better you can wield it. Isis Chen, a UX Researcher on the Xbox Research team, is earlier in career, describing AI as the 24/7 mentor she never had. And Hilary Dwyer, a Principal UX Researcher, is the senior strategist reframing the whole role around judgment and foresight.

The pattern that emerges is subtler than youth vs experience. AI fluency is fast to acquire; experience is slow. What AI accelerates is the gathering of knowledge – what it can’t shortcut is the judgment that only years of building real products with real people produce. On that, the junior and the two seniors agree.