Some work directs algorithmic systems. Some work is directed by them. Same job title can sit on either side. The economic and career stakes are enormous - and most workers haven't noticed which side they're on.
Algorithmic management used to be a niche topic about Uber drivers and Amazon warehouse workers. AI made it the default modality of work - and now applies to knowledge workers too.
Above-the-algorithm work uses AI to gain leverage. Below-the-algorithm work gets allocated, scored, and managed by AI. The two sides live in different economies.
Above the algorithm means you use or build algorithmic systems to gain leverage. You direct them. Your decisions shape what the system does. Examples: the engineer who chose which model to deploy, the marketer who uses AI to amplify their judgement, the product manager who decides what an AI feature should do.
Below the algorithm means the system directs you. It allocates your tasks, scores your performance, fires you when scores drop. Examples: ride-hailing drivers, warehouse pickers, increasingly customer-support agents and content moderators. Most consequentially, AI is now pushing this dynamic up the value chain - into legal discovery, junior consulting, and entry-level analysis.
The split runs through every industry AI touches. The same job title - "analyst," "associate," "designer" - can sit on either side depending on how the org has architected the work. Where you sit determines whether AI is a multiplier on your work or a manager of it.
A perfectly engineered system where humans direct nothing.
Amazon's fulfilment centres are the cleanest example. AI decides which item a picker fetches next, what route they walk, how long each task should take. Performance is tracked to the second. Underperformers are flagged automatically.
The workers are skilled and the system is impressive. But notice who's above and who's below: the engineers who designed the system are above the algorithm. The pickers are below it. The economic value flows up. The managerial discretion flows up. The pricing power flows up. Picker wages are bounded by what the algorithm decides the task is worth.
The same pattern is appearing in knowledge work. AI paralegal tools allocate document-review work to junior lawyers and score their accuracy. AI sales tools assign leads to SDRs and track their pipeline. The juniors are being moved below the algorithm without anyone calling it that.
The line between above and below isn't about skill or education. It's about who is the directing intelligence in the workflow.
A senior engineer at a startup who chooses which AI assistant to use, evaluates its output, decides what to ship - above the algorithm. Their wage is rising.
A junior engineer whose tickets are allocated by an AI-managed sprint planner, whose code is graded against an AI baseline, whose performance is benchmarked against AI output - below the algorithm. Their wage is compressing. Same job title. Two different economic realities.
A senior agent who designs scripts, escalation paths, and tone-of-voice guides for the AI agents that handle 80% of tickets - above the algorithm. Few of them, well-paid.
A line agent whose remaining 20% of tickets is routed by an AI triage system, whose response time is scored, whose tone is monitored - below the algorithm. Many of them, wages bounded by the system.
Five quick questions. Pick the answer that sounds more like your work. Honest matters more than aspirational - the value is in seeing where you sit.
When something at work gets prioritised, the priority typically comes from:
When you make a judgement call that contradicts the system's suggestion:
Your performance is measured by:
When a new AI feature ships in your tools, your first reaction is:
If you stopped showing up tomorrow, the system would:
Above-the-algorithm work is where the leverage lives. Below-the-algorithm work is where the wages stop growing.
An advisory engagement that redesigns the operating model with explicit choices about which roles direct the algorithms and which roles are directed by them - and the long-term economic consequences of each choice.