AI is doing your Data Science – now what?
Time saved by AI is real — but it raises new questions
ChatGPT has greatly reduced my labor time. I think this trend will converge similarly across all Data Scientists. Let’s say 60-80% of time saved - meaning only 40-20% of the original time is now needed to produce the same output. Most of the time saved is coming from technical tasks like coding, but they can also come in the form of faster ideation.
In most environments, only this visible labor time will be acknowledged, because it typically leads to tangible outcomes. As a result, managers/stakeholders will begin to question how the rest of the time is being spent. They’ll start looking to redeploy that time or consider labor reduction. This is classic resource optimization. And it will probably happen to all the functions, not just Data Science.
A manager’s moment — but not everyone wants that
I hate to say it, but this is an epic time to become a people manager. The core responsibility is resource allocation or orchestration, and with AI freeing up significant bandwidth, the ROI can be massive. But let’s say you don’t want to be a manager. How do you optimize as a Data Scientist in this new environment? And how should employers and leaders adapt their expectations?
The topology has shifted
Think of Data Science as a function that has inputs and outputs. The outputs are clearly some measure of business value. What about the inputs? The skills we typically associate with data scientists — mathematical thinking, product intuition, technical fluency — are not inputs. They are characteristics of how the function behaves. In mathematical terms, they define the topology of the function. Every company will have its own Data Science function in a topology increasingly influenced by AI.
The actual inputs to the function are things like business context, data, and goals — what gets transformed into value through the process.
So how do you optimize business value from Data Science? You always had and will continue to have two levers:
Optimize the inputs — data quality, meaningful hypotheses, judgement, etc
Optimize the function itself — stronger mathematical skills, deeper product intuition, etc
Both ultimately lead to some weighted average of two outcomes: solving more problems of the same complexity, or solving problems of greater complexity.
AI is doing some Data Science — but human judgement elevates it
With AI entering the picture, we are working within a new topology — one that reshapes the path from question to insight. It also enables more people to perform data science. But I’d still bet that a team of trained Data Scientists will outperform a random group of professionals navigating this landscape. The marginal advantage may shrink over time as the models improve and non-experts become more capable. But unless AI reaches full functional abstraction, traditional Data Scientists will retain a meaningful advantage.
And given Data Science’s reliance on context, interpretation, and causal ambiguity, full abstraction is a hard target. It can happen — but we’re not there yet.
So what elevates Data Science in this new topology?
More people can do the job - but value still concentrates
To reiterate, it is now easier for more people to perform the duties of a Data Scientist or the same Data Scientists to do more of the similar tasks. But the incremental value — the edge over your business competitor — will still come from those with traditional grounding and experience in the field. They will either do more or take on more challenging problems.
With that, here are a few recommendations to both consumers of the Data Science function and the Data Scientists themselves to maximize value.
For consumers of Data Science
Ask more open-ended and ambiguous questions. AI has widened the search space — but use that breadth purposefully.
Avoid systematically reclaiming the time AI has saved. Let your teams use it for deeper, exploratory work.
Hire for stronger foundational Data Science skills, not weaker. AI has raised the baseline — raise your bar.
When in doubt, trust your Data Scientist over some AI model. If you're hiring well and holding high bars on accountability, you won’t go wrong betting on human judgement. Models are powerful — but they don’t fully understand context, nuance, or organizational dynamics the way a thoughtful Data Scientist does.
For Data Scientists
Use AI if you aren’t doing so. Let it assist you, challenge your assumptions, and be a thought partner.
Build trust — it will matter more as AI levels the field.
Protect the time AI saves. The real unlock isn’t in saving time — it’s in reinvesting that time into solving more difficult, more ambiguous problems that grows you.
Expect to be challenged more. Use it to clarify your thinking, improve your communication, and grow your edge.
Even with AI, don’t overlook mathematical grounding. It supports better judgment.
AI is changing the role, but a lot of the ways to maximize value are still grounded in traditional principles.
Thanks for reading!
