As AI moves from experimentation into everyday products and working life, the questions surrounding it are becoming increasingly human. Drawing on conversations and presentations captured through AIBA, Clare English, Marketing Director at intO, explores an emerging shift from concerns about what AI can do to more complex questions of trust, judgement, expertise, culture and what we should choose to keep human. As intO prepares for AIBA 2026, these ideas will form the starting point for a new research project tracking how practitioner thinking about AI is evolving, and what successful human-AI collaboration might ultimately require.
For the past three years, I’ve led intO’s involvement in co-organising AIBA, the AI Business Advantage conference in Katowice, Poland, working particularly closely with the speakers and contributors to our Research & Innovation programme. It’s given me an unusual vantage point on the AI conversation: not only hearing what people working across Product, Research, Design and Innovation choose to talk about, but helping them develop the perspectives, research and experiences they want to bring to the stage.
Over the past week, I’m been looking back over those conversations to understand how much they have changed.
When we began our involvement as co-organisers in 2024, much of the conversation around AI still focused on the technology itself: what it could do, how quickly it would improve and what possibilities it might unlock. Alongside the excitement were increasingly urgent questions about bias, representation, access, explainability and the implications of building technologies for a global population on data and assumptions that reflected only a fraction of it.
None of those questions has gone away. But as AI has moved rapidly from experimentation into products, organisations and everyday working life, another set of questions has come to the fore. They are less about the capabilities of the technology and increasingly about the relationship we’re building with it: what we should automate and what should remain human; when we should trust AI to act on our behalf; what happens to expertise when machines can produce expert-looking work; and how people develop judgement if AI removes some of the work through which judgement has traditionally been acquired.
Perhaps the biggest question of all is becoming: as AI gets better at doing things for us, what do we actually want humans to bring to the relationship?
It’s a question I’ve found myself returning to repeatedly as we’ve shaped this year’s AIBA Research & Innovation programme around Designing Successful Human-AI Collaboration. And it has also made me curious about the archive we’ve accumulated along the way. We haven’t yet completed the analysis, so I don’t want to impose a neat story on it prematurely, but looking back at the interviews, presentations and discussions we’ve captured so far, an intriguing evolution in the AIBA conversation is beginning to emerge.
Alongside the conference programme in 2024, intO interviewed people working across AI, Product and Design about the development of equitable AI. Those conversations repeatedly brought us back to bias, access, representation, explainability and human oversight, but what interested me then, and interests me even more in retrospect, was the broader idea sitting underneath them.
Many of the problems we were attributing to AI weren’t necessarily new problems. They were longstanding problems of technology, organisations and design that AI suddenly had the capacity to reproduce faster and at far greater scale.
Culture made this particularly visible. One interviewee, drawing on experience in autonomous vehicles, challenged the assumption that a single model of human behaviour could work everywhere. A system developed around how people behave on the streets of San Francisco cannot simply assume that those behaviours will hold in Mumbai. Technology may travel globally, but human behaviour doesn’t become universal as a result.
Others raised related concerns about Western-centric training data, unequal access to technology and products created by relatively narrow groups of people for populations whose lives might look very different from their own. Taken together, those conversations suggested something that still feels highly relevant today: AI doesn’t create our organisational and cultural blind spots, but it can make them scalable.
When we returned to Katowice in 2025, the material being brought to the Research & Innovation stage was notably more concerned with what happens when people actually encounter AI in products, professions and everyday decisions.
At Booking.com, for example, UX research was exploring what makes people trust AI when planning travel. What emerged was far more nuanced than a simple measure of whether users did or didn’t “trust AI”. Trust changed according to context and stakes. During the inspirational stages of planning a holiday, relevance and the freedom to explore mattered, whereas accuracy became much more critical as someone moved towards spending significant money on a flight or accommodation. A system acknowledging its limitations could build rather than diminish trust, while something as apparently straightforward as ignoring a user’s stated budget could rapidly undermine it.
In financial services, we heard a related perspective on the need to build confidence gradually before asking people to allow AI to perform increasingly consequential actions. Getting a song recommendation wrong is irritating; transferring money to the wrong person is something else entirely. Rather than thinking about trust as a feature we can simply add to an AI product, these conversations suggested that it is better understood as a relationship, shaped by context, consequence and human expectation.
That distinction feels increasingly important as AI moves from systems that answer questions towards systems capable of acting on our behalf.
One of the presentations that stayed with me most from AIBA 2025 came from Amsterdam’s UX Research community, because it captured how quickly AI was beginning to change not only what researchers did, but how they thought about the value of their profession.
Their earliest community discussions had been practical, focusing on prompting, tools and the points in the research process where AI might help. Before long, however, the conversation had become much more existential. If AI can write discussion guides, summarise interviews, identify themes and generate polished reports, what distinguishes an experienced researcher from somebody capable of prompting an AI system effectively?
The obvious answer is the quality of the thinking behind the output, but that introduces a harder problem. Expertise develops through practice. Researchers learn to recognise weak evidence, ask better questions, spot contradictions and understand what matters partly because they have spent years doing difficult cognitive work themselves. If AI increasingly provides a shortcut through that work, how do less experienced practitioners acquire the judgement they will eventually need to recognise when its answers are wrong?
That question reaches well beyond UX Research. If AI removes some of the work through which judgement is learned, where does the next generation of human judgement come from?
Another idea I found particularly interesting in the 2025 programme was a challenge to one of the most deeply embedded assumptions in digital product development: that identifying friction and removing it is almost invariably a good thing.
AI vastly increases our ability to do that, but it also makes the question of whether we should more important.
Research presented by Microsoft looked not simply at pain in people’s working lives, but at joy, care, meaning and psychological wellbeing. One deceptively simple example concerned performance reviews. Writing them takes managers time and AI could undoubtedly make that process faster, but spending time reflecting on another person’s contribution is also an act of management. Remove the effort and we may inadvertently remove some of the value embedded within it.
Research with warehouse workers revealed something similar. Asking not only about people’s tasks and problems but about what they found meaningful in their work surfaced dimensions of their experience that a narrow efficiency lens would never have revealed.
For me, this points towards a much richer way of thinking about automation. Instead of asking only what AI can do, or even where we need to retain a “human in the loop”, perhaps we should be asking what the human is bringing to the loop, and whether it is something we want to preserve.
Context is one of the threads I’ve noticed running through these conversations from the beginning. It shapes whether someone trusts an AI recommendation, which activities they are comfortable delegating, how they interpret an interaction and what they consider useful, appropriate or valuable. It becomes particularly consequential when the products we’re designing are intended to operate globally.
This is also where the AIBA conversation intersects closely with our work at intO. Through our Equitable AI Framework and our research with Product and UX teams across markets, we’ve been exploring the ways AI can accelerate innovation while simultaneously accelerating the assumptions, biases and blind spots embedded within the systems and organisations deploying it.
There’s an interesting paradox here. We might have imagined that increasingly intelligent technology would make local knowledge less important, yet I increasingly wonder whether the opposite is true. As AI enables products to become more personalised, conversational and autonomous, understanding the human context in which those products operate becomes more consequential, because trust, privacy and expectations around money, health, work and communication aren’t culturally universal. Even our understanding of “value” changes according to context.
The more capable AI becomes, the more consequential context may become.
This October, I’ll be returning to Katowice with the intO team and our cohort of speakers for AIBA 2026, where our Research & Innovation programme will explore Designing Successful Human-AI Collaboration.
As we’ve been developing the programme and working with this year’s speakers, it has increasingly felt like the natural next stage of the conversation. AI capability will continue to advance, but for Product, UX, Research and Innovation teams, the more difficult challenge may be deciding how that capability should fit into people’s lives and work.
We’ll be exploring where human judgement creates value alongside automation; what creates trust as AI moves from recommending towards acting; how successful human-AI collaboration might differ across markets and cultures; and what else we should optimise for if efficiency is no longer the constraint it once was.
I don’t expect AIBA to provide a single answer to those questions. What interests me is bringing together people who are encountering them from very different positions, and seeing where their experiences converge and where they don’t.
It’s also prompted us to begin a deeper research project at intO. Across the AIBA conferences we’ve co-organised, we’ve accumulated interviews, presentations and discussions from people working at the intersection of AI, Product, Research, Design and Innovation, and we’re now going back through that material more systematically.
We want to understand what practitioners were concerned about in 2024, which questions persisted into 2025, what became visible only as AI entered everyday products and working practices, and what people are wrestling with now. We haven’t finished that analysis, and there is more material still to add, so the observations I’ve made here are a starting point rather than its conclusion.
At AIBA 2026, we’ll add another layer by speaking directly to this year’s speaker cohort about the emerging challenges of human-AI collaboration. After the conference, we’ll bring those conversations together with the earlier material in a new intO whitepaper that will explore how the AIBA conversation has evolved across three extraordinary years of AI development, where consensus appears to be emerging, where important tensions remain unresolved, and what all of this might tell us about what comes next.
I’ll be sharing more from the AIBA archive and introducing some of the people contributing to this year’s conversation in the run-up to Katowice. Following AIBA 2026, we’ll publish the full whitepaper.
Clare English
Marketing Director, intO
Get in touch with clare@studio-into.com | Connect with me on LinkedIn
Following conversations with more than twenty senior Product and UX Research leaders, intO Founder and CEO Joanna Brassett is taking the conversation to the Bay Area. This September, intO will bring together senior leaders to explore how AI, continuous discovery and shifting responsibilities are reshaping the relationship between Product and UX Research.
Product organisations have never had more access to customer insight. So why are better-informed teams finding decisions harder to make? Nitika Wahi explores one of the biggest shifts reshaping Product and UX Research today.
