Data Science Is Dead Again - Simon Oster, Arcadis
June 25, 2026
Members in the News: Simon Oster, Arcadis. Every few months, a new job role is declared dead. Even data science. But in reality, if you think about it, agentic AI has a surprisingly familiar machine learning mindset. My opinion: data science is sexy again.
Data science is not “dead,” but it has changed significantly. In the past, the focus was on building and training models from scratch. Today, the work is more about designing systems around existing AI models (such as LLMs).
The core process remains similar:
- Then: model → data → train → evaluate → iterate
- Now: model → context → tools → evaluate → iterate
The main difference lies in how systems are improved (no longer through training, but through prompts, context, tools, and evaluation).
The text highlights that:
- Good AI solutions require substantial engineering (it’s not just “adding an LLM”).
- Reliability, monitoring, and evaluation are crucial.
- Agent-based systems (AI that can perform actions) are more complex than they appear.
- Observability (insight into what the system is doing) is essential to ensure quality.
Conclusion:
Data science is still very much alive, but it is shifting toward system design, software engineering, and building AI agents. The required skills are evolving, but the role remains important.
Share this post!