'/%3E%3C/svg%3E){kind=small}
It is the only Artificial Intelligence craft that US federal statistics record as a standalone occupation: a median pay of 112,590 dollars and a projected growth of thirty-four per cent over the decade. And in Europe? The needs analysis puts it at the top of the most sought-after roles. But behind the most overused title of the decade, what counts remains the method: a portrait of the AI Data Scientist.
When a craft enters the official labour statistics, it stops being a fad. The Data Scientist entered them long ago: the Bureau of Labor Statistics (BLS), the US federal office for labour statistics, documents a median annual pay of 112,590 dollars in 2024 and a projected employment growth of thirty-four per cent between 2024 and 2034, among the highest of all recorded occupations, summarising the function thus: “Data scientists use analytical tools and techniques to extract meaningful insights from data” (BLS).
The Occupational Information Network (O*NET) of the Department of Labor details the tasks: “Develop and implement a set of techniques or analytics applications to transform raw data into meaningful information […] Apply data mining, data modeling, natural language processing, and machine learning”. In the sense that concerns us here, the AI Data Scientist combines statistical, programming and domain competences to develop predictive models, tending to the quality of the datasets and communicating the results to decision-makers.
Method before algorithm
The substance of the craft lies not in the tools, which the repertoire lists in abundance – from analysis languages to machine-learning libraries – but in four disciplines of method.
The first is problem framing: translating a business question into a well-posed statistical one, choosing the relevant metrics, and recognising when the available data cannot answer. The quality of an analysis is decided here, before any algorithm.
The second is rigour: correct sampling, out-of-sample validation, control of spurious correlations and of information leakage between training and testing, and the quantification of uncertainty. Without this safeguard, models predict the past and fail the future.
The third is responsibility: verifying the representativeness of the data, measuring the effects of the models on the various groups of data subjects, and documenting limits and conditions of use. This is the point at which data science meets Regulation (EU) 2016/679 (the GDPR) and, for systems falling within the high-risk categories, the data-governance requirements of Regulation (EU) 2024/1689 (the AI Act).
The fourth is communication: honest visualisations, the explicit statement of assumptions and margins of error, and the courage to tell decision-makers even what they would rather not hear. An incommunicable result is unusable; a poorly communicated one is worse.
And in Europe? At the top of the needs list
European demand is no less explicit than the American. The AI-skills needs analysis of the EU co-funded Artificial Intelligence Skills Alliance (ARISA) opens its list of the most needed roles precisely with this profile: “The AI practitioners’ roles that are needed most are data scientists, data engineers and especially machine learning engineers”. And the overall pool in which this demand moves is expanding sharply: according to the Union’s statistical office, “in 2025, more than 10 million persons were employed as ICT specialists across the European Union (EU)”, equal to 5.0 per cent of total employment (Eurostat).
The Italian benchmark
For the most inflated professional title of the decade, Italy now has an antidote to dilution: the UNI 11621-8:2026 standard defines the profile with competences set out according to the methodology of the European e-Competence Framework, the UNI EN 16234-1 standard, and anyone wishing to have those competences attested may undergo assessment by an accredited body under the UNI CEI EN ISO/IEC 17024 standard, within the framework of Accredia Information Circular DC No. 21/2026.
Conclusions
The AI Data Scientist stands at the crossroads of statistics, computing and domain knowledge: Washington certifies the demand with federal statistics, Europe places it at the top of its needs, and Italy has made its competences verifiable. There remains the point that no statistic can guarantee: the difference between data science and its marketing is made, always, by method.
In the light of the above, one wonders how many, among the many who today carry this title, would withstand a third-party assessment of their competences; the market, now, has the tool to ask for it.
