What does data governance mean? The vague meaning has consequences for ethics and AI. Chris Wiggins and Alison Powell argue it’s time to articulate a vision of governance that centres social participation, shared decision-making and equitable distribution of benefits and risks.

The word “governance” has become a Rorschach test for our hopes and anxieties about data and power. We can track the term’s drift from state-level responsibility to corporate performance. The governance “driftword” lets its meaning stay vague, causing consequences for ethics, AI and even statistics.

As data has evolved from methods for tabulating the greatness of states in the 19th century to algorithms empowering our personal, professional, and political reality, the term “governance” has drifted alongside it, with meanings shaped based on who controls data access and policy, and towards whose goals. Data governance matters because data systems increasingly impact people’s lives and the functions of social institutions. More robust mechanisms for meaningful public decision making are needed, alongside more meaningful interpretations of governance within government and industry.

In industry, governance can be reduced to compliance and risk management. In academia, it expands to encompass social participation and decision-making. In government, it increasingly narrows to focus on “efficiency” and “growth”, terms that mask deeper questions about who benefits from and who bears the risks of technological change.

Recent shifts in UK policy illustrate this tension. While public discourse celebrates AI adoption and digital transformation, civil servants grapple with reduced capacity and mandated use of AI tools without clear applicability. The gap between rhetoric and reality widens: growth-focused policies advance, while discussions of social impacts recede. “Growth and security” narratives compete with those advancing participatory governance. Decisions about the adoption of data-driven systems are often made without public consultation – because of a perception that people aren’t interested in or don’t understand data-driven systems, even when these systems influence their lives.

The drift of key terms

“Governance” is not the only term that has experienced semantic drift in the data world. Even “statistics,” “AI,” and “data science” have all shifted dramatically since their introduction. As Chris and his co-author Matthew Jones demonstrate, data science and modern AI resulted from a continuous progression of mathematical developments empowered by algorithms acting on data. The meaning of these terms has been contested, expanded and narrowed through decades of technical, social and political evolution.

No term has experienced more dramatic semantic drift over the past decade than “ethics.” The book “How Data Happened” illuminates that data has long been used as both “a tool and a weapon in arguing for what is true, as well as a means of rearranging or defending power”. Between 2017-2023, “ethics” became a capacious luggage term carrying our collective hopes, dreams and fears as data-empowered algorithms increasingly shaped our personal, professional and political realities.

People are interested, engaged and motivated to make different kinds of decisions about and with their data, as demonstrated by Alison’s experience with communities in East Walworth, London. But this kind of ethics in practice or “from the bottom up” disappears when “data ethics” becomes associated with corporate governance boards or complex system design features.

The contested terrain

The evolution of “ethics” in data contexts reveals how terminology becomes a battleground. Despite the many contributions to society that have emerged from data science, benefits often accrue to those defending power, rather than empowering the defenceless. What began as principled discussions of accountability has often been co-opted into corporate “ethics washing,” substituting for meaningful regulation or structural change. These performative gestures can distort policy-making processes so that as Alison and coauthor Fenwick Mckelvey point out, their “drama” becomes a focal point. This can undermine efforts to develop effective and enforceable principles.

When Chris presented the historical arc of data ethics at our panel, he traced this evolution spanning centuries. From early statistical collection used to develop and apply theories of social systems to today’s AI systems, accountability and ethical standards consistently lag behind technological capabilities. The “Battle for Data Ethics” chapter in the book demonstrates how ethics itself became a contested term, with different stakeholders attempting to define it in ways that served their interests.

Towards meaningful governance

Venture capital has shaped not just technology adoption but our very conception of value in data systems. Just as Uber’s subsidised rides once created an illusion of sustainable, cheap transport, today’s AI systems often obscure their true costs –  both economic and social. As danah boyd noted about “How Data Happened,” the book places “data in context so that we can see the values, politics and controversies that shape our present reality”..

The challenge that our conversation addressed is moving beyond critique to articulation of specific solutions. How do we address the real costs of AI and data systems? How do we confront issues like the radiologist shortage in Europe, which exemplifies how AI adoption can create unexpected labour market effects? And most importantly, how do we reclaim “governance” as a term that centres social participation rather than technical compliance?

Dr. Seeta Peña Gangadharan, the chair of our panel, argues that both friction and alignment will be central to moving beyond privacy and “fairness” frameworks, towards sustaining just and democratic values in a datafied society. Her examination of governance strategies reveals how refusal, alongside antitrust regulations and technical solutions, might help address power asymmetries in data-driven contexts.

Our panel explored multiple perspectives on these questions:

• UK business community frameworks, including insights from Erin Young’s work at the Institute of Directors on innovation and technology policy

• regulatory and compliance approaches, informed by research on responsible AI governance

• social participation models, building on Alison’s research on civic action in smart cities and her work directing the JUST AI network

• infrastructure and computational power considerations, including analysis of who benefits from current technological arrangements

By examining these diverse viewpoints, we can begin to understand how “governance” might be reclaimed as a meaningful term rather than a “driftword” that obscures more than it reveals. Dr Young’s research on gender gaps in AI and data science provides a crucial lens for understanding how structural inequalities in the field affect the technologies being built. As her work demonstrates, addressing representation matters not just for equity but for creating systems that serve diverse populations.

Conclusion

“How Data Happened” argues that by understanding the trajectory of data – where it has been and where it might go – we can “bend it to ends that we collectively choose, with intentionality and purpose”. The same might be said for the language we use to discuss data governance.

In Alison’s work on technologies in the public interest, she argues that the design and context of our systems reflect the values we prioritise. This research examines how discourse shapes not just our understanding of technology but its material implementation – sometimes creating unintentional marginalisation. Initiatives like Alison’s “data walking,”  demonstrate how participatory methods can reframe technical discussions to centre community needs.

We have an opportunity to define our terms clearly and intentionally. Rather than allowing “governance” to remain a Rorschach test for conflicting interests, we can articulate a vision of governance that centres social participation, shared decision-making and equitable distribution of both benefits and risks.

The words we use to frame these discussions matter. By acknowledging the drift in terminology and deliberately choosing our language, we can shape the discourse around data governance in ways that expand rather than limit our collective imagination.

Sign up for our weekly newsletter here.

• This blog post represents the views of its author(s), not the position of LSE Business Review or the London School of Economics and Political Science.
• Featured image provided by Shutterstock.
• When you leave a comment, you’re agreeing to our Comment Policy.

Print Friendly