Career Stages in the AI Era
The traditional career ladder was built on the assumption that the first few years of a person's career would be spent doing well-defined tasks under direction, gradually building the pattern recognition that lets them take on harder problems. AI has collapsed that assumption. Most of the well-defined-task work that used to be Level 1 of the ladder is now done faster and more cheaply by AI than by a junior practitioner.
This page is about what that means for how Product Teams design career paths and grow people.
Four cognitive levels of work
A person's career stage is best identified not by their title or years of experience, but by the question they spend most of their working time answering.
| Level | Question | Value the person brings |
|---|---|---|
| 1 | "Is it done?" | Speed and reliability on specific tasks. |
| 2 | "Is it solved the right way?" | Judgment to choose the best methods and trade-offs across full solutions. |
| 3 | "Is this the right problem?" | Cross-functional systems thinking to identify the real problem to solve. |
| 4 | "Is this the right direction?" | Calibrated decision-making under uncertainty about what outcomes to pursue. |
These are not seniority levels in the traditional sense. They describe the cognitive work the person is being asked to do. Two people with identical titles and tenure can be operating at different levels, and frequently are.
The vanishing floor: Level 1 is being automated
For decades, the first one or two years of a career were spent almost entirely at Level 1: doing the thing under direction, learning the ropes, building reps. AI is now an excellent Level 1 employee. It produces outputs faster than a junior, and the quality bar continues to rise.
The implication is that Level 1 has effectively vanished as a sustainable career stage. People can no longer spend years at it building pattern recognition. To be employable at all, new entrants need to move into Level 2 quickly: directing the systems that produce the work rather than producing the work themselves.
This creates a real problem for how Product Teams develop people. The repetition that used to build judgment is gone, but judgment is still needed. The team has to develop it deliberately rather than letting it accumulate as a side-effect of doing tasks.
Level 2 as the new entry point
Level 2 is no longer something practitioners arrive at after a few years. It is the floor.
A practitioner operating at Level 2 needs to know the different solution options for a problem and the trade-offs of each. When a stakeholder says "improve onboarding," the Level 2 practitioner can think through tutorial flows, contextual tips, empty-state designs, in-product video, and explain when each fits.
The practical question for Product Teams: how does someone reach Level 2 without first spending two years at Level 1? Three approaches that work:
- Studying principles deliberately. What used to be absorbed implicitly through repetition now has to be learned explicitly through reading, watching senior practitioners work, and being walked through their reasoning.
- Shipping side projects. End-to-end ownership of small projects exposes the practitioner to the trade-offs they will not see on a single Stream Team.
- Apprenticeship and pairing with Level 3+ practitioners. Teaming and apprenticeship programs are how the senior practitioner's reasoning gets transmitted, not the senior's outputs.
Level 3: where humans still win
Level 3 practitioners stop accepting the problem statement as given. When a stakeholder says "we need a mobile app," the Level 2 practitioner asks "Flutter or React Native?" The Level 3 practitioner asks "why do we think an app solves the retention problem?"
This requires seeing across functional boundaries. A design decision affects engineering. A product choice ripples into sales cycles. A pricing change impacts customer success. AI excels at individual tasks but currently struggles to connect insights across disciplines coherently. This is where human leverage is concentrating.
The practitioner shape that fits Level 3 is comb-shaped: moderate depth across multiple areas, with the ability to connect them, rather than the classic T-shape with one deep specialism. The trap to avoid is staying inside your functional home. If you only ever see the design perspective, you will only ever see part of the picture.
Level 4: setting the frame
At Level 4, the practitioner is no longer diagnosing problems within a frame someone else set. They are deciding what outcomes the organisation should pursue at all.
These decisions happen under uncertainty. Data is incomplete. Causality is unclear. Feedback arrives late and filtered. The work is making evidence-informed bets and creating enough shared belief in the team to test them.
The trap at Level 4 is that the practitioner's influence changes shape. Casual suggestions now consume organisational resources. Half-formed ideas become directives. The practitioner needs to know when to use their weight and when to hold back, and they need people around them who will tell them the truth even when it's uncomfortable.
Implications for Product Teams
Three practical consequences for how the Product Team designs career paths and supports people:
- Level 1 hiring no longer makes sense as a stand-alone strategy. Hiring people whose only output is Level 1 work means hiring people who cannot do anything AI can't already do. Either invest in pulling them up to Level 2 quickly through structured apprenticeship, or don't hire at that level at all.
- The performance levels need to be rewritten around the question the person can answer, not the tasks they can complete. A senior IC who is still asking "is it done?" is competing with a machine that doesn't sleep.
- Apprenticeship and teaming become the primary mechanism for moving people up the ladder. The repetition path is gone. The visible-senior-judgment path is what's left.
What this doesn't mean
Two reactions to the AI collapse of Level 1 that are tempting but wrong:
- "Just hire fewer people and rely on AI." This works for output in the short term and starves the senior pipeline in the medium term. The Level 3 and Level 4 people of five years from now have to come from somewhere.
- "Re-create Level 1 work as 'AI training'." Watching someone supervise an AI is not the same skill-building loop as doing the work. Apprenticeship works because the apprentice has agency over the work, not because they are passive observers.
The honest answer is that growing senior practitioners is now a deliberate activity that costs time and attention. Product Teams that invest in it produce the people the organisation will need; teams that don't will hit a ceiling within a few years as their existing seniors move on.