Sunil Venkataram
Back to garden
•
🌰 seedling
Job Dimensionality - Why Low-Task Jobs Face the Highest Automation Risk
type/noteailaboreconomics

Job Dimensionality — Why Low-Task Jobs Face the Highest Automation Risk

Two reasons dimensionality matters

Reason 1: Conditional displacement. If a job has 20 tasks and one gets automated, a human still does the other 19. The job survives. If a job has one task and it gets automated, the job is gone. The probability of full displacement given partial automation scales inversely with the number of tasks.

Reason 2: Firm incentives to invest. Automating a task costs real money (buying software, onboarding, integrating with existing systems). A firm weighs the investment against the payoff. If automating the last remaining task eliminates a position and its entire wage bill, the return on investment is high. If automating one task out of twenty just makes the worker slightly more efficient, the return is lower. Firms will work harder and spend more to automate low-dimensional jobs.

This second reason is the one most analyses miss. Standard exposure indices treat all “unexposed” tasks as equally protective. They are not. In a low-dimensional job, the remaining tasks do not keep the worker employed once the core tasks are automated. In a high-dimensional job, they do.

The trucking case

Roughly 3 million Americans drive trucks for a living. Long-haul trucking is dominated by a single core function: moving the truck safely from point A to point B. Logistics, loading, and unloading are handled by others. If autonomous driving becomes reliable on long-haul routes, the job is not augmented. It is eliminated.

Companies like Aurora Innovation and Kodiak Robotics are already running commercial autonomous trucking on constrained routes. The firm incentive to invest is strong: eliminating the driver eliminates the largest per-mile cost and removes the constraint of mandatory rest stops and shift limits.

Warehousing tells a similar story. Picking, packing, sorting, and pallet movement are narrow tasks increasingly performed by machines. “Dark warehouses” that run around the clock with minimal human labor already operate abroad.

The consulting counter-example

A management consultant’s job combines research, data analysis, client communication, presentation design, strategic reasoning, team coordination, and relationship management. That is at least seven or eight distinct complementary tasks. AI might automate the data analysis and slide deck creation, but the consultant still handles everything else. The O-ring focus effect kicks in: the consultant spends freed-up time on client relationships and strategic judgment, raising quality across the board. See O-Ring Production and AI Automation - Why Partial Automation Can Raise Wages.

Medicine works the same way. A physician’s job spans diagnosis, treatment planning, patient communication, care coordination, documentation, and clinical judgment. Automating documentation (the most common current use of medical AI) frees time for the rest. The job has too many dimensions for any single automation to eliminate it.

The prediction this generates

Even if a job is not currently “exposed” to AI in the standard sense (AI cannot yet do the tasks involved), if the job is low-dimensional and the technology is approaching the capability threshold, that job should be considered at high risk. The firm incentive to close the gap is strong.

Conversely, a job with high exposure scores but many complementary tasks may be the safest kind of job to hold. High exposure in a high-dimensional job means AI augments many tasks, triggering the focus effect and potentially raising wages.

The relevant measure is not average task exposure. It is the structure of bottlenecks and whether automation reshapes worker time around them or eliminates the position.

Backlinks (4)
Connected Notes