Most of us are now familiar with AI entering the workplace through tools like Copilot, Grammarly and ChatGPT. For many, the benefits are clear. But alongside those benefits comes a level of uncertainty: about shifting role boundaries, changing ways of working, and what these technologies may mean in the longer term.

Despite this uneasy undercurrent, in many organisations, honest conversations haven’t happened. There’s been little direct discussion about what AI means for people, roles and the future of work. In that gap, uncertainty tends to fill the space, and for many employees, anxiety quietly follows.

The anxiety of ambiguity

While it would be easy to dismiss this as simply a fear of technology, that would miss the point. What many organisations are experiencing is something more precise. The dominant emotion in AI adoption isn’t fear of technology, it’s the anxiety of ambiguity.

The pressure to adopt AI and not fall behind means that AI lands (often quickly) before anyone has meaningfully worked out what it means for the people doing the work. The technology moves faster than understanding, communication and clear direction, and in that gap, people naturally fill the space with their own inferences.

These fears are exacerbated by a fundamental speed mismatch. AI capability moves at software speed; organisational culture moves at human speed. The gap between the two is where much of the anxiety sits. Expecting one to keep pace with the other by default isn’t a realistic strategy.

Yet in many organisations, there is often a natural assumption that introducing a new tool will quickly lead to embedding it in day-to-day work, and that access will translate into adoption and then comfort, which rarely holds true in practice. The result is not resistance so much as a sustained state of uncertainty, where people are working with tools whose implications for their roles are still being understood.

The dilution of expertise

The market research industry is no stranger to a digital transition, but this wave feels particularly stark in its potential impact. Roles that were once considered highly skilled have become partially commoditised, and capabilities that took years to build are now available at the click of a template.

When working in quant and analytics, technical skill – the ability to build and interpret statistical models, write analysis code, manage complex datasets – has long functioned as a marker of credibility and capability. AI compresses the value of that expertise quickly and visibly. Tasks that once required fluency in R, that required genuine statistical understanding, can now be initiated by someone with neither.

While this creates real opportunities for upskilling, particularly for those previously constrained by time or resources, it also needs to be understood as more than a practical shift. It is an identity shift. When what you know is central to how you are seen professionally, widespread access to that knowledge changes more than job descriptions. It destabilises the professional self.

Opportunity or unsettlement

The question this raises of whether researchers are ultimately technicians or strategists is not a new one, but AI makes it urgent. It also makes it uncomfortable, because most roles in the industry have historically required both, and have been valued for both. Removing that balance forces a rapid shift in how many researchers understand their own value within the industry.

This naturally leads to questions around roles, and whether AI is likely to bring greater clarity to them or instead unsettle them further. It has the potential for both, but the reality depends on the level of meaningful integration. The optimal version of the story is that AI handles the mechanical work, on the high-value interpretive work they likely entered research to do. This is not the easiest path.

Given the differences in pace outlined earlier, it would require sustained, deliberate effort to keep that balance intact as tools evolve and expectations shift, so in practice, the picture is considerably messier. AI’s speed originally promised freedom for deeper thinking, but too often, it just creates more work. Compressed timelines mean output expands while understanding doesn’t. Researchers who push back to protect quality, context, and evidential rigour risk being cast as blockers rather than guardians.

Ensuring good integration

What helps is less glamorous than most AI strategies suggest. It is not about tool access, training courses, or transformation roadmaps. It is about explicit, role-level conversation: what is AI being asked to do here, what does that change, what stays human, and who has the standing to shape those answers? This is a mutually beneficial goal as this is often something clients seek clarity on too, to be reassured that their investments are not being handed over to an AI.

The organisations navigating this well have moved beyond the efficiency framing
altogether. They’ve recognised that ‘doing more with less’ is a promise that arrives with an implicit threat attached, and they’ve chosen a more honest account of what they’re actually asking their people to become. Employees hear: you will be freed up. They suspect: your role will be reduced. In that gap, efficiency stops feeling like empowerment and starts feeling like exposure.

The pace of culture

Whether most organisations will get there is a different question. The tools are evolving faster than the cultures that hold them, and the emotional labour of navigating that gap continues to fall, quietly and unevenly, on the people doing the work. Organisations with more agile management styles are better placed to absorb the transition, giving teams the space to adapt as the landscape settles.

Where culture is slower to shift, uncertainty lingers and clarity takes longer to follow. By and large, the pace and impact of AI will be shaped as much by organisational mindset as by the technology itself.

Chrissie Tarbotton is research director at Boxclever, and Shane Hanson is director of insights and innovation at Panasonic Consumer Electronics, Europe

The full Delphi report, ‘Who owns understanding? How AI is reshaping value, expertise and accountability in research’ is accessible via the MRS website.