New LSE/HKU research argues AI splits 'weak-bundle' jobs into cheaper tasks while leaving 'strong-bundle' roles intact but augmented.
Economists Luis Garicano (LSE), Jin Li, and Yanhui Wu (University of Hong Kong) published a paper reframing AI's labor market impact. Rather than mass job elimination, they argue AI 'unbundles' roles — automating separable tasks in weak-bundle jobs (hollowing them out) while augmenting strong-bundle jobs where tasks are interdependent. The paper directly challenges forecasts like the 10.4 million US jobs lost by 2030 figure, arguing task-level AI exposure doesn't map cleanly to job-level displacement.
Developers sit in strong-bundle territory: debugging, architecture, code review, and client communication are interdependent enough that AI augments rather than replaces. But the products developers build — automation pipelines, AI agents, task-specific tools — are precisely the mechanisms that unbundle weak-bundle jobs at scale. Understanding this distinction changes how you think about which workflows to automate and what residual human oversight is legally or practically necessary.
Audit one of your current automation projects: list its tasks and ask whether the human role is a 'bundle' (interdependent decisions) or a checklist. If it's a checklist, you're building something that replaces a role, not augments it — price and position accordingly.
Suggested exercise · not independently tested. Access requirements and costs may vary.
Go to claude.ai and open a new conversation
Paste this prompt: 'I'm building an AI tool that automates [describe your tool, e.g. invoice processing for accounts payable teams]. Using the Garicano weak-bundle vs strong-bundle framework, classify whether this role is a weak or strong bundle, list the tasks AI will replace vs augment, and flag any legal/compliance reasons the human must stay in the loop.'
Review the output to identify whether your product hollows out a role or enhances it — this affects your liability exposure and sales narrative
Tags