The Argument That Still Holds
Five years ago, I wrote a piece arguing that generalists are more valuable than specialists in data teams. My reasoning was simple: while specialists excel at solving well-defined problems, generalists define the problems in the first place. They navigate the ambiguity, connect the dots, and bring in specialists only when needed.
Re-reading that article today, I still agree with almost everything. But one subtle shift has occurred that changes the entire landscape: AI has become the new specialist.
AI now handles many of the tasks we once relied on human specialists for—deep expertise, clear briefs, well-defined instructions. And it does so faster, without fatigue, and at scale. This doesn't diminish the generalist; it amplifies them.
The Wicked Learning Environment Never Left
We don't operate in neat, closed systems. We operate in what David Epstein calls wicked learning environments—settings where rules are unclear, feedback is delayed or misleading, and patterns don't repeat consistently. In these environments, you can do the "right" thing and still fail, or the wrong thing and appear successful.
AI doesn't remove this ambiguity. If anything, it amplifies it. When answers come faster and look more convincing, the risk of confidently solving the wrong problem only increases. The generalist's ability to question the problem itself becomes even more critical.
This is the core insight: the real challenge is not solving problems—it's knowing which problems are worth solving. And that requires range.
Hyper-Specialization Is Being Pushed to the Edge
Back then, I argued that access to information (Stack Overflow, blogs, documentation) reduced the need for deep specialization. A capable generalist could figure things out quickly enough.
Today, that dynamic has shifted dramatically. AI doesn't just help you find the answer—it gives you a working answer. It curates, synthesizes, compares, and presents information in an instant.
The result? Hyper-specialization isn't disappearing, but it's being pushed closer to the edge. Generalists are now empowered to go much further before needing specialist input. You can build an end-to-end data pipeline, deploy a model, and set up monitoring—all without a dedicated DevOps engineer or ML specialist on speed dial.
Coordination Cost Is Still the Real Killer
One of my strongest points then, and even more true today: the cost of coordination in organizations is often underestimated. Generalists reduce this cost by eliminating unnecessary handoffs. They need the mandate to make decisions and cut through the management overhead.
Jeff Bezos popularized the "two-pizza team" rule: teams small enough to be fed with two pizzas. In today's world, we're heading toward one-pizza teams. Not because the work is simpler, but because generalists are more capable, and AI fills many specialist gaps, resulting in fewer handoffs required.
The Business Problem Hasn't Changed
Strip everything back, and the core questions remain the same:
- How do we grow revenue?
- How do we retain customers?
- How do we operate more efficiently?
The tooling has evolved. The methods are more sophisticated. But the underlying problems are unchanged. Businesses don't care whether the solution involves a cutting-edge agentic model or a well-placed SQL query. They care about results.
As a proteome-scale protein structure prediction guide might require both deep bioinformatics knowledge and practical engineering skills, the same principle applies: the value lies in connecting the problem to the right solution, not in mastering every tool.
What Changed? (And What Didn't)
| Aspect | Then (2020) | Now (2025) |
|---|---|---|
| Specialist role | Human expert | AI as default specialist |
| Generalist scope | Connector between specialists | Connector + deep executor with AI |
| Coordination need | Two-pizza teams | One-pizza teams |
| Problem definition | Critical | Even more critical (AI amplifies ambiguity) |
| Information access | Available | Curated and synthesized instantly |
The New Generalist: Amplified, Not Replaced
The key shift is this: generalists are no longer just connectors between specialists. They are the ones navigating environments where the problem is unclear, the signals are noisy, and the path forward isn't obvious.
They connect not only people, but capabilities—deciding when to trust intuition, when to rely on experience, and when to bring in an on-demand specialist, human or AI.
Their range is now amplified, capable of executing much deeper work themselves. Not because the world became simpler, but because they still operate well in complexity, with AI as their ever-available specialist layer.
Limitations and Caveats
This doesn't mean specialists are obsolete. Deep expertise is still critical for edge cases, novel research, and high-stakes decisions where AI's confidence intervals are wide. The generalist's advantage is in breadth, but depth still wins when the problem is well-defined and the stakes are high.
Next Steps for Your Career
If you're a data professional wondering where to focus:
- Double down on problem-definition skills. Learn to ask better questions, not just find faster answers.
- Build your AI toolkit. Learn how to leverage AI for specialized tasks—code generation, data analysis, documentation—so you can execute deeper work without waiting for a specialist.
- Develop your communication and coordination skills. The generalist's superpower is translating between domains.
- Stay curious. The moment you stop exploring adjacent fields is the moment your range starts shrinking.
I'm looking forward to reflecting on this again in five years. The only certainty is that the landscape will shift again—and the generalist will be there to navigate it.
This article is a reflection on the original piece published on Towards Data Science.
