Description

This Corpus note captures Abi’s biggest takeaway from Jim Sterne’s Experimentation Elite talk: AI may not kill expertise, but it could damage the learning paths that create it.

The piece argues that AI is increasingly able to take on the “bottom rung” of professional work: data collection, dashboards, first-pass analysis, test ideas and reporting. But that work was never just grunt work. It was apprenticeship in ugly trousers: the place where juniors learned how to notice patterns, make mistakes, build judgement and understand what quality looks like.

Abi is careful not to make this a nostalgia argument. Some skills can disappear without tragedy. The real question is which losses matter. Using the cabinet-maker metaphor, the piece distinguishes between speed and judgement: AI can make the cabinet faster, but it cannot reliably tell whether it is heirloom or Ikea. A dashboard is not insight. A test idea is not strategy. A summary is not understanding. A recommendation is not responsibility.

The core argument is that AI should help people practise thinking, not skip it. Used well, it can challenge assumptions, preserve context, explain history and force better questions. Used badly, it turns humans into arms and legs for the machine. The future skill is judgement, not because humans are magical, but because humans are accountable.

Topics

• AI and the automation of junior work

• Why “grunt work” often functions as apprenticeship

• The difference between preserving old tasks and preserving learning paths

• Speed vs quality in AI-assisted work

• Output vs judgement

• Why more output can mean faster nonsense

• The human smell test

• Why “technically correct” is not the same as right

• AI as a tool for practising thinking

• Judgement, accountability and expertise in AI-shaped work

Best for

UX researchers, experimentation teams, analysts, digital strategists, AI leads, managers, mentors, and anyone replacing junior work with AI without asking how people will become senior later

Background

This piece is part of Abi’s Corpus work on AI, judgement and better digital decision-making. It extends the Corpus argument beyond websites and discovery into how teams learn, reason and protect expertise in AI-shaped workflows.

It also connects to the wider Corpus view that more output is not the same as better understanding. Whether the subject is UX, experimentation, AI or upstream optimisation, the critical question remains the same: what evidence is being used, what assumptions are being skipped, and who is responsible for deciding whether the answer is actually any good?