How Agentic AI Is Reshaping the Skills That Define a Data Professional

Abstract

As AI becomes more autonomous in analyzing data, the skills required to become a data professional are evolving. As a prequel to the previous generations of tools that helped a human execute every step, more agentic systems are increasingly able to perform multiple steps of analytical tasks, from data preparation to model building to the final report, with less involvement from a human. This paper discusses the implications of this change for the skills which make up professional value in data work. It suggests that the skills that once constituted competence are being translated to problem framing and judgement, oversight of automated systems, and cross-human–machine collaboration as agentic systems take over the technical execution of competence. Based on published research since 2015, covering automated machine learning, the automation–augmentation relationship, algorithmic control, and the division of cognitive labor between the human and the machine, the paper classifies these skills into those that automated agentic systems make less salient and those that they make more salient. It argues that the ability to state a problem well, to determine whether and when an automated output can be trusted, and to monitor systems that now execute rather than just advise is increasingly a professional skill. The paper also challenges the concept of professional education, which has been historically built around mastery of technical execution as the indicator of expertise, and proposes that it is time to move away from this conception towards a focus on judgment and oversight, which are still human. It's not about what the data professional can do with data; it's about what the data professional can do with data that machines can do.