AI Needs More Data Science Thank You Might Think

What is the role of data scientists in the booming field of AI that seems to be dominated by engineers?

While AI engineers are building powerful systems, data scientists bring some unique skills that are often overlooked — yet absolutely crucial. And most of these skills are complementary to AI engineering, not in competition with it. I see four key areas where data scientists naturally shine:

Data Understanding & EDAI’ve had numerous experiences working alongside ML/AI engineers who built AI products without carefully looking at the data even once! Data scientists, on the other hand, are painfully aware of the GIGO principle (Garbage-In-Garbage-Out). Data Scientists know that data isn’t just raw input — it has quirks, missing pieces, and often carries subtle biases. Deep exploratory data analysis (EDA)* is not just a phase; it’s a mindset that shapes everything downstream.
ExperimentationIn general, we as a community need to think more deeply about how to bring scientific thinking into AI through hypothesis-driven development, rigorous experimentation, and continuous learning from data. Framing hypotheses, designing experiments, testing, and iterating — all of this is second nature to data scientists. A scientific approach ensures that our models are not just outputs of engineering pipelines, but are thoughtfully designed and stress-tested.
Measurement & EvaluationKnowing what to measure, how to measure it, and what the results actually mean is a key differentiator. Data scientists bring statistical rigor and domain intuition that helps ensure AI systems are not just technically impressive but also reliable, fair, and aligned with their intended outcomes. Data Scientists understand that precision and recall aren’t just metrics — they’re trade-offs with real-world consequences.
ExplainabilityTurning black boxes into glass boxes is where data scientists excel. They’re trained to ask “why” and to communicate it in clear, accessible ways. Whether it's for debugging, stakeholder trust, or regulatory requirements, the ability to explain model behavior is not optional — it's essential.

All of these traits aren’t just nice-to-haves. They’re fundamental to building responsible, robust, and human-aligned AI.

Let’s not forget: AI is not just built — it’s measured, evaluated, improved, and explained. And that’s exactly where data scientists can make a big difference!

What is the role of data scientists in the booming field of AI that seems to be dominated by engineers?

Related Posts

8 Key Principles for Applied AI

The Coding Evolution of a Data Scientist: 20 Years, 4 Eras