Beyond Adoption: Measuring AI's real-world impact on people, work and communities

AI is being adopted. But is it creating impact?

Is your use of AI actually working, for your people, for the communities around you? Most organisations investing heavily in AI cannot answer that question with confidence. They can measure deployment, adoption and usage. What they often struggle to measure is whether AI is improving people's work, benefiting communities or creating meaningful value for the organisation itself.

And that matters. Without evidence of real-world impact, organisations risk mistaking adoption for success. They may scale initiatives that show high adoption but deliver limited value, while overlooking unintended consequences and emerging risks. Increasingly, organisations are being asked not only how they use AI, but what difference it is making for people, communities and society.

This paper explores how organisations can answer that question with evidence, and more specifically how to move beyond adoption metrics to understand AI's real-world impacts on people, work and communities.

AI adoption is growing, but trust remains fragile

Ipsos and Google's AI Works America research highlights the gap between AI usage and real-world outcomes. While 40% of U.S. employees reported using AI at work, only 5% qualified as AI Fluent. Those workers reported significantly greater productivity gains, time savings and career benefits, suggesting that usage alone does not explain the outcomes people experience.

At the same time, confidence in AI remains uneven. Two in three American workers already believe AI will make their working life worse rather than better, while AI usage remains concentrated among higher-income, college-educated, white-collar employees. Among those using AI at work, the greatest gains accrue to those who have become most fluent in it. France follows a similar pattern: despite growing adoption, confidence in AI remains mixed. Only 38% of people view AI positively overall, 60% say it makes them nervous, and just 26% trust companies to protect their personal data when using AI.

These findings raise questions not only about who benefits from AI, but also about who trusts it. Organisations need to understand who benefits, who may be left behind, what concerns people have and what conditions help AI deliver positive outcomes.

The stakes extend beyond the workplace. AI infrastructure, including data centres, can create economic value while raising concerns about energy and water use, local employment and community impact. Whether these developments are considered acceptable depends not only on their economic benefits, but also on how their effects are experienced and understood by local communities. Organisations need evidence to understand both the outcomes and the concerns that shape trust and social acceptance. This is becoming a critical component of responsible AI, helping organisations move beyond deployment to understand real-world consequences and ensure benefits are shared fairly.

Start with the problem, not the technology

AI touches almost every area of business and society, from recruitment and supply chains to healthcare, education, environmental monitoring and customer service. Yet organisations are grappling with very different questions depending on the decisions they need to make.

What narrows it usefully is not the technology but the business question behind it. A recruitment algorithm that skews hiring away from workers over 50 is not primarily a technology problem. It is a fairness problem. A data centre being built next to a residential community is not primarily a technology problem. It is a community-impact problem. The changing content of jobs is not primarily a technology problem. It is a workforce-transition problem.

Framing the question correctly determines what needs to be measured and what evidence organisations need to make better decisions.

Getting to the root: adoption data is not impact data

Most existing measures of AI in organisations focus on adoption and attitudes: usage trackers, engagement measures, sentiment surveys and self-reported productivity claims. These help organisations understand whether people are using AI and how they feel about it, but they cannot establish what has actually changed in people's work, communities or daily lives.

Self-reported metrics in particular have limitations. Asking a business leader whether AI has reduced headcount, or an employee whether AI has made them more effective in their role, captures perceptions rather than necessarily demonstrating outcomes. These perceptions may reflect assumptions about AI rather than evidence of what has changed.

What has AI actually changed in people’s work, communities and opportunities, and who is benefiting or being disadvantaged? Where are these developments leading, what unintended consequences are emerging, and what actions would improve outcomes for people, communities and organisations? Answering these questions requires combining multiple sources of evidence: quantitative data that shows what has changed, alongside direct conversations and observations that explain how these changes are experienced in practice. Looking at both provides a much fuller picture than either alone.

While AI presents a new context, the measurement challenge is not entirely new. Ipsos has previously applied similar impact measurement approaches to areas such as community development, sustainability and human rights, helping organisations understand what changed, for whom, and with what consequences.

So what? Moving from insight to action

For an organisation ready to move past adoption metrics and find out what its AI deployments are actually doing to people, four steps get this started:

1. Identify where evaluation matters most. Map AI use cases and prioritise those with the greatest potential consequences for business performance, workers or communities, such as recruitment systems, workforce management tools or major infrastructure investments.
2. Measure beyond usage. Combine operational indicators with independent quantitative and qualitative research to understand what has changed, for whom and why. Include outcomes such as trust, fairness, workforce experience and unintended consequences.
3. Turn evidence into action. Use the findings to identify where AI is delivering value, where risks or inequalities are emerging, and what needs to change to improve outcomes for people, communities and the organisation.

Organisations do not need to have all the answers about AI today. But as AI becomes embedded in recruitment, workforce management, customer experience and infrastructure, understanding its effects will become increasingly important.

That means looking beyond adoption figures and productivity claims to understand what is happening in practice: how people’s work is changing, how communities are affected, who benefits and where concerns remain. Without that understanding, organisations risk making decisions based on an incomplete picture of AI’s value and consequences.