What if the most efficient team isn't the best team?

One of the more plausible consequences of AI is smaller teams.

If a group of people can produce significantly more with the same effort, organisations will eventually ask whether the same output could be produced by fewer people.

That is a reasonable question.

But it assumes that output is the only thing a team produces.

Teams also produce capability.

People learn by watching colleagues work, asking questions that would never justify a formal meeting, seeing mistakes corrected, disagreeing with somebody more experienced and gradually absorbing how decisions get made.

Teams produce challenge.

A good colleague notices the assumption nobody else questioned. They bring a different experience, remember something the group has forgotten or simply refuse to accept the first plausible answer.

Teams produce resilience.

Knowledge is distributed. Somebody can be absent without the work becoming impossible. Several people understand why a decision was made rather than only what the decision was.

Teams also produce things that are less comfortable to put into a productivity calculation: belonging, identity, confidence, friendship, shared memory and the feeling of achieving something with other people.

None of this means smaller teams are bad.

In many cases they may be better. Communication can become easier. Accountability can become clearer. People can have greater autonomy. AI may remove coordination overhead as well as production work.

The problem is that the benefits of smaller teams and the costs of losing parts of a team are measured differently.

Productivity is visible.

If five people can now produce what ten produced before, the arithmetic is easy to see.

The learning that no longer happens between the missing five and the remaining five is much harder to count.

So is the challenge that never occurs, the organisational memory that gradually narrows, or the future leader who never develops because the layer of work through which they would have learned has disappeared.

What if the most efficient team isn’t the best team?

This connects directly to the apprenticeship problem, but it is broader than apprenticeship.

The question is not only how junior people become experienced. It is what a team is for once production capacity is no longer its only scarce contribution.

AI may force organisations to become more explicit about that.

If a team exists only to aggregate enough human effort to produce an output, then dramatic productivity gains should change its size.

But if the team also exists to develop people, challenge decisions, retain knowledge, provide resilience and create the social conditions in which people do good work, those functions still need to exist even when fewer hands are required for production.

Perhaps the right question is not simply how many people the work now requires.

It is which capabilities the organisation still needs the team to create.