When AI takes over research, administration and first drafts, leaders need to preserve the learning those tasks used to provide. Junior people still need opportunities to make a decision, explain their reasoning, hear a more experienced perspective and learn from the outcome.
In a recent LinkedIn discussion, I asked what happens to the first rung of the career ladder when AI does work that people previously learned through. The question for leaders is practical: where will judgement develop inside the redesigned role?
What disappears when the first draft disappears?
A first draft can reveal more than writing ability. It shows what someone noticed, which evidence they trusted, what they left out and where their understanding is incomplete. A senior colleague can use that unfinished thinking to teach.
If the only thing a manager sees is a polished AI-assisted answer, those learning opportunities become harder to spot. The output may improve while the person’s reasoning remains untested. Faster production alone does not tell us whether someone is becoming more capable of making the next decision.
Keep an independent view in the process
One useful approach is to ask a junior colleague to form a short initial view before comparing it with an AI response. This need not mean repeating every administrative task manually. It means preserving a moment in which the person decides what they think and why.
The comparison then becomes a conversation. Where do the two answers agree? Where do they differ? Which assumptions need checking? What evidence would change the recommendation?
In my replies to the LinkedIn discussion, I emphasised making room for people to explain where the model changed their mind and why. Revising a view in response to better evidence matters as much as defending the original view.
Make the reasoning visible
For a team experimenting with this approach, a short review can start with four prompts:
- Your decision: What would you recommend before seeing the model’s answer?
- Your evidence: Which facts support that recommendation, and which still need checking?
- Your revision: What did the AI response change, and why did you accept or reject that change?
- Your next check: What outcome would tell us that the decision needs revisiting?
These are suggested prompts for a learning conversation, not a claim that one process fits every role. The amount of independent work and review should reflect the person’s experience and the consequences of the decision.
Protect time for experienced judgement
AI can make output faster without making coaching automatic. A more experienced colleague still needs to question assumptions, explain trade-offs and help someone connect a decision to its outcome.
The leadership commitment is to make time for those conversations. If every minute saved becomes another production target, the team may lose the space in which people learn. A useful review therefore considers both what was delivered and what the person can now explain or do more independently.
A question to take into your next role review
Choose one task that AI has changed in your team. What did a junior person used to learn while doing it? Identify where that learning will happen now, who will support it and how you will know it is developing.
That is a more useful starting point than assuming either that every old task must be preserved or that better output automatically means better judgement. The work can change. The responsibility to develop people remains.
Continue the discussion
This article develops the ideas in my LinkedIn post and follow-up discussion on AI and the career ladder. For related reading, explore five ways to develop as a leader, lasting influence in leadership and the Technology & AI collection.