The ambiguity of 'productivity'

11 Nov, 2026

Did you have a productive day?

We generally describe ourselves as having had a productive day when we do all those things which we planned to do. They didn't take longer than expected. There were no unexpected difficulties or challenges and there were no unexpected interruptions. That is even compatible with a little bit of procrastination. On a very productive day, we may do things we planned more quickly than expected and either move on to something we planned for another day or take some time off.

We may be unproductive because we have made an error about how long a task would take or the difficulty involved in completing it. With tasks that are novel, or familiar tasks in new contexts, this is to be expected. Unproductivity is problematic when we succumb to excessive procrastination, or we are unnecessarily interrupted and distracted by the demands of others, or some system or tool we rely upon fails.

We also think that personal productivity tools can increase the chance of being productive in this way. Generally, such tools are means of organizing tasks, time and information so that we do not suffer other forms of interruption or disturbance of our productivity. At the most basic level, a good to do list allows one to know what to move on to next without having to stop and spend time thinking about that. Filing systems and databases allow us to access information without delay. Good organization of meetings and effective channels of communication also reduce the chance of distraction and delay.

Productivity in this sense is achieved by having well founded expectations of what can be achieved in any given period of time and having tools, systems and approaches which reduce the risk of tasks taking longer than expected. We can call this personal productivity or effectiveness.

Economic productivity

Notice that in the sense of productivity we have just been talking about, someone who chooses to slowly and carefully hand craft an item, whether it be code, pots or words, may be consistently productive. They may consistently achieve in a day what they expected to achieve in that day, even though someone else, who perhaps automates some part of the process, or takes bigger risks on quality, might produce more in the same amount of time.

Is the person who takes a week to hand knit a sweater less productive than the person who takes a couple of hours with a knitting machine? Yes, but only if we use the word 'productive' to mean something completely different. Even if hand knitted sweaters sell for three or four times the price, the person using a knitting machine has added more economic value in any given period of time. They are more economically productive. They may even maintain this greater economic productivity while being less personally productive or effective.

So what?

It is important to labour this ambiguity because failure to keep it clearly in mind allows for exploitation of workers. In most forms of employment there are incentives and rewards for being personally productive or effective, these range from simple bonuses through to elaborate career structures. If capitalism works, and it is far from obvious that it does,¹ these incentives will also reward economic productivity.

Once this structure is in place, it is easy for employers to increase economic productivity, which benefits them and not the employee, by reducing the amount of time a task is expected to take. If this is done by introducing targets, expected product in a given time period, which are impossible to achieve, or the achievement of which will result in clear and direct harm to the employee, then employers will meet resistance. Employees will go off sick, or leave, or, even worse, form unions.

So capitalism incentivises the employer to find ways of increasing economic productivity which the employees will accept. And equivocation on the term 'productivity' is a very effective means of doing this.

Labor saving devices

The introduction of AI, specifically LLM-based tools, into clerical and managerial jobs is a very clear example of this. If the volume of work, level of interruption and poor systems and processes are such that a productive day is very rare, offering an employee a tool which increases the chance of a productive day is very attractive. It is a productivity tool in the sense of one that increases the chance of completing the tasks assigned within the expected time.

However, for the employer, the sole function is to increase economic productivity.² We can see this because we could make each employee more personally productive, more effective, by simply having more employees engaged. This will keep the number of tasks assigned to any individual sufficiently low to allow them to be productive in respect of completing those tasks with space for interruptions, systems failures, etc. However, employing more people reduces the economic value created by each person. It is less economically productive.

It feels good to be effective, to be on top of your job, and bad to be consistently, day after day, unproductive. Before the advent of AI, this led to middle managers consistently arguing for more team members in order to do the work their team was expected to do. Now they get told to use LLMs. These are sold as a solution to challenges with personal productivity and thus packaged as a benefit to the employee, making it easier for them to do their job well (and receive the rewards for so doing).

However, unlike previous productivity tools, these do not function by removing hindrances to productivity, but by decreasing the expected time for completion of the task. We need to see them like that to know why we should resist them.