A graduate joins a team, ready to contribute. Before they have finished understanding the assignment, an AI tool has produced the brief, drafted the proposal and summarised the customer conversation. The documents look competent. The work moves faster. But something important remains unresolved: where will that graduate learn to make a good decision?
My answer is not to protect repetitive work. It is to redesign apprenticeship around evidence, explanation, responsibility and feedback. If AI produces more of the first draft, leaders must become more deliberate about how people develop judgment.
Remove busywork, not the learning inside it
Nobody needs to spend a career formatting slides to earn the right to think. Some work was always repetitive, and some apprenticeship models confused endurance with development. We should welcome tools that remove unnecessary effort. We should be more careful about removing the learning hidden inside it.
Writing a brief can teach someone to distinguish a customer’s request from the problem underneath it. Preparing a recommendation can expose missing evidence. Summarising a conversation can reveal that two people left with different expectations. If AI produces the document, those learning opportunities do not automatically survive. Leaders have to design them back into the work.
Better output is not the same as independent judgment
The research gives us reasons for both optimism and caution. In a customer-support study, researchers found that AI assistance particularly benefited less-experienced workers, with suggestive evidence of learning. That matters: AI can help people access expertise earlier. It does not follow that stronger work with assistance is the same as independent judgment across professions. Read Generative AI at Work.
A separate study of knowledge workers found an association between greater confidence in AI and lower self-reported critical-thinking effort. It did not establish that AI makes people less capable or causes long-term skill loss. For me, its practical implication is narrower: checking, questioning and taking responsibility must become explicit expectations, rather than activities we assume happen behind a polished answer. Read the Microsoft Research publication.
Give the junior colleague something real to own
Consider an illustrative assignment: a junior colleague preparing a product recommendation. Let AI generate the first version. Then ask the colleague to challenge it:
- Which assumption is weakest?
- What evidence comes directly from customers?
- What would make the recommendation wrong?
- Which alternative did they reject, and why?
The assignment is no longer simply to deliver a document. It is to develop and defend a position.
Next, give that colleague a real conversation. Ask them to explain the recommendation to the customer, with a senior person present to support them, not rescue every difficult moment. A customer’s unexpected question can reveal a gap that an internal review misses. Afterwards, discuss what changed, what remained uncertain and what they would do differently.
Responsibility should grow within clear boundaries. A junior colleague can own a small decision, know when to escalate and receive feedback after the outcome becomes visible. The senior person still carries oversight. But oversight should not mean quietly rewriting everything. If all the reasoning stays with the manager, the junior learns approval-seeking rather than judgment.
AI adoption and Saudi capability belong together
This matters for young Saudis entering work. Vision 2030’s Human Capability Development Program emphasises lifelong learning and skills aligned with labour-market needs. These are programme objectives, not evidence that any particular employer has achieved them. My reading is that employers have a practical responsibility: not merely to recruit potential, but to create work through which that potential develops.
For a boutique AI and product studio such as Siyada Tech, I see this as a useful leadership standard, not a claim about existing practice. The question is not only what a tool can deliver for a customer. It is whether the team can explain its choices, recognise its limits and become more capable through using it. That is the human side of our boutique approach.
Preserve the path to judgment
I would judge an AI-enabled team by whether its juniors can increasingly explain, challenge and own the work. Speed matters. So does knowing when the fast answer is wrong.
When we redesign a role around AI, we should be able to explain how someone will grow through it. We do not need to preserve busywork. We need to preserve the path to judgment.