Every Artificial Intelligence application we use today was born, years ago, in a scientific paper. The AI Research Scientist is the profile that generates that knowledge: it formulates theories, conducts reproducible experiments, publishes and submits to peer review. The US repertoires record it precisely; for Europe, which has set itself the target of twenty million digital specialists, the real question is another: will it be able to keep it?
There is a long timescale, in Artificial Intelligence, that the market cycles do not tell: the architectures that today underpin generative systems were born in the laboratory long before they became products. That time belongs to the AI Research Scientist, the figure devoted to exploring the frontiers of the discipline to generate new knowledge and theoretical paradigms, ensuring methodological rigour, reproducibility and responsibility.
The craft, in the official US repertoires
The Occupational Information Network (O*NET) of the US Department of Labor describes the function thus: “Conduct research into fundamental computer and information science as theorists, designers, or inventors. Develop solutions to problems in the field of computer hardware and software”. The Bureau of Labor Statistics (BLS), in its Occupational Outlook Handbook, lists among the typical tasks: “Explore problems in computing and develop theories and models to address those problems” and “Write papers for publication and present research findings at conferences”. The product of the work, then, is not a system in operation: it is validated knowledge, destined to be taken apart, verified and surpassed by the scientific community.
Method, reproducibility, responsibility
Three disciplines define the quality of this profile. The first is problem formulation: identifying the open questions that matter, telling the incremental from the foundational, translating intuitions into testable hypotheses; the directions most explored today – from learning with reduced supervision to architectures that integrate symbolic reasoning and neural networks, up to multimodal systems – are born of well-posed questions before they are born of well-executed experiments. The second is the experimental method: comparisons with honest baselines, the publication of code and experimental conditions, the acceptance of falsifiability; in a field exposed to recurring clamour, reproducibility is the line that separates science from announcement. The third is responsibility: assessing the implications of the paradigms that open up, from fairness to explainability and on to computational sustainability, documenting the limits together with the results, because research on AI is not ethically neutral and those who conduct it are the first safeguard of its consequences.
Europe, between declared targets and global competition
On the demand side, the European context is defined by a legal target and by a tension. The target lies in Decision (EU) 2022/2481 on the Digital Decade: “the number of ICT specialists employed in the Union is at least 20 million” by 2030; and the growth is under way, if it is true that, according to the Union’s statistical office, “from 2015 to 2025, the number of ICT specialists in the EU increased by 59.4%, more than 6 times the increase (9.8%) in total employment” (Eurostat). The tension lies in the global competition for frontier researchers, contended by laboratories and firms on every continent: for Europe, the match is not only to train this figure, but to offer it the conditions – of resources, computing infrastructure and scientific freedom – to stay.
The Italian benchmark
The UNI 11621-8:2026 standard also includes the AI Research Scientist among the professional role profiles of AI, with competences set out according to the methodology of the European e-Competence Framework, the UNI EN 16234-1 standard, and with the possibility of assessment and certification of professional competences at bodies accredited under the UNI CEI EN ISO/IEC 17024 standard, within the framework of Accredia Information Circular DC No. 21/2026. That the most academic profile of the whole chain is brought back to verifiable competences speaks to the ambition of the framework: to give the professional ecosystem of AI, from research to operation, a common language.
Conclusions
In the beginning, in Artificial Intelligence, there is always research; the applications, the products and even the standards come afterwards, and live off it. The AI Research Scientist is the custodian of that principle: their quality is measured in reproducibility and scientific honesty, their presence is measured in a territory’s capacity to generate the future rather than import it.
In the light of the above, one wonders whether Europe, and Italy with it, will be able to offer their frontier researchers sufficient reasons to stay; every scientific paper written elsewhere by a talent who left from here is an application, a business and a piece of technological sovereignty that will be born elsewhere.
