AI and data: AI is transforming roles, but above all it is strengthening accountability
In many companies, artificial intelligence is now being integrated into reporting, financial analysis, and operational management tools. Conversational agents, natural-language queries, automatic generation of KPIs: AI is speeding up access to information and simplifying tasks that have historically been technical in the data space.
But behind this promise of analytical autonomy, one reality stands out: AI does not replace data expertise; it redefines its scope and accountability.
So the question is not only technological. It is above all organisational.
Can we really trust the results produced by AI?
Trust in AI does not rest on its infallibility, but on rigorous oversight: audits, testing and governance to uncover and control its weaknesses—where human error, more visible, remains easier to attribute.
-
Scepticism towards AI often concerns its reliability.
An error generated by an automated system seems more serious than a human error: faster, potentially affecting considerable volumes, it appears systemic and difficult to attribute.
Yet in practice, the impact and the means of preventing errors are not so different from human action.
The difference lies elsewhere: an employee can be identified, challenged and held accountable. An AI, on the other hand, does not legally bear fault.
This implies a shift in posture. The tool must be supervised, audited, tested—not because it is intrinsically less reliable, but because its ability to produce convincing results can mask underlying weaknesses.
Trust does not disappear with AI.
It must be structured.
A transformation of roles rather than a disappearance
AI automates part of the technical tasks, refocusing data experts on their core work: defining KPIs, documentation, validating results, and accountability for the generated outputs.
Contrary to some common misconceptions, artificial intelligence does not signal the end of data roles.
Historically, data projects relied heavily on strong individual technical skills:
-
Development
-
Modelling
-
Building complex queries
AI agents now handle part of this technical exploration. But that does not reduce the value of experts—quite the opposite.
Their role is evolving towards:
-
Proper definition of KPIs
-
Understanding and formalising business rules
-
Validating calculations and generated results
Value is shifting from technical production to the guarantee of consistency.
This evolution directly affects finance, management control, marketing, and supply chain functions: all roles built on analysis and operational steering are impacted. AI speeds up processing, but paradoxically strengthens the need for human expertise.
A sense of déjà vu: from the Big Data wave to the AI era
The arrival of AI in data raises challenges similar to those of Big Data: data quality, formalising rules and governance—now with greater urgency to resolve them in order to fully harness its potential.
-
This upheaval echoes that of Big Data in the 2010s. Even then, companies were asking:
Where can we find the relevant data?
How can we use it effectively?
For which concrete use cases?
Issues around data quality, poorly defined business rules, or insufficient documentation have never completely disappeared.
The difference today lies in the level of exposure. By opening up new use cases, AI makes these weaknesses immediately visible to everyone.
An approximate KPI or an implicit rule that went unnoticed in manual processing becomes a source of inconsistency amplified by automation.
AI does not create these problems. It reveals them.
Concrete use cases… provided you have solid foundations
AI excels at document exploration and report generation, but its effectiveness relies on structured data, clear business rules, and appropriate governance.
In the field, the most effective use cases remain pragmatic:
-
Ad-hoc exploration of existing data
-
Accelerated document search
-
Generating summary analyses or reports without specific development
-
Development and analysis assistance for non-technical profiles
Some platforms now integrate these capabilities directly at the heart of data environments, making natural-language interaction and the automation of analytical tasks easier.
But these use cases only create value if the foundations are solid:
-
Reliable, structured data
-
Clearly formalised business rules
-
Defined governance
-
Established validation processes
Without these elements, AI can produce results that appear consistent… but are fragile in their interpretation.
Towards new accountability for data roles
AI’s reliability depends on human control: its ability to convince must not obscure the need for rigorous audits and process traceability to ensure credibility.
-
Artificial intelligence does not devalue data roles: it repositions them.
The expert is no longer limited to writing queries or developing models. They become the guarantor of:
Consistency of KPIs,
Traceability of calculations,
The control framework,
Accountability for the results used in decision-making.
As AI spreads across organisations, the central question is no longer only “what can it do?”, but “who is able to validate it?”.
The companies that will derive the most value from AI will be those that accept the need to strengthen their data fundamentals before automating its use.
This article is inspired by an op-ed published by Camille Maurice, Data & Analytics Director at MeltOne, in the newspaper Informatiques News in 2026
Let’s discuss your challenges and our solutions
Let’s talk about your project and discover how MeltOne can turn your challenges into concrete, high-performing solutions.
