AI as a teammate. Messier than expected.

The promise of partnership

Most people don't want to work for a robot, but many of us are okay working with one.

The promise of AI was always that it would take care of the dull bits. The reports that write themselves. The spreadsheet that finally understands context. The admin that quietly does its job so you can do yours. The reality has been slower and messier, but the idea still holds: technology should make work feel more human, not less.

The reality check

The trouble is that teamwork doesn't come naturally to machines. They don't read the room, share intent, or know when to keep quiet. Collaboration requires empathy, and AI has none. What it does have is potential. It's a tireless assistant that never runs out of coffee or ideas (however creative), but it still needs direction.

Recent research from the MIT Centre for Collective Intelligence found that human–AI combinations don't automatically outperform AI or humans alone. On average, collaboration worked best when each did what they do best: humans brought creativity, judgment and context; AI brought speed and precision. In short, collaboration is a relationship you build.

The apprenticeship model

In the past year, I've seen teams use AI like a colleague they're training. They check its work, correct its tone, teach it the company style and gradually trust it with more. Despite the language of automation, it can feel more like a slightly frustrating apprenticeship. The machine learns from the people, and the people learn what the machine can actually do.

As Harvard Business Review calls it, this is collaborative intelligence. A cycle where humans explain, train, and correct, while AI amplifies what humans do best. The workflow evolves: the system drafts, the human decides, and both improve over time.

The hybrid workforce

We have entered what Forbes describes as the rise of a hybrid workforce, where teams consist of both humans and AI agents.

Leaders will soon manage not only human employees but also AI colleagues; digital teammates that never tire but still need guidance. Some companies already list AI agents on their org charts. Others use AI orchestrators, or digital middle managers that coordinate specialised systems.

It sounds futuristic, but it's just the next phase of teamwork. HR meets IT; culture meets code. Collaboration is no longer about where we work; it's about how we think and simulate thinking together.

What makes collaboration work

The same principles that make human teams thrive also apply when AI joins the group:

1. Shared goals. Everyone — human or not — needs to know what success looks like.
2. Complementary strengths. Let people handle context and creativity; let AI handle repetition and scale.
3. Transparent feedback. Good teammates show their reasoning. So should algorithms.
4. Psychological safety. Teams learn faster when it's safe to question the system.
5. Continuous learning. Collaboration isn't static; it improves with every iteration.

That balance matters. Give AI too much responsibility, and you lose control. Give too little and you miss the point. The best setups feel like a partnership where the system drafts, the human decides, and both improve over time. It's not perfect, but neither are we.

The human advantage

The teams that thrive with AI are the ones that stay curious. They experiment, laugh at the messiness of it all, and move on. They frame AI as just another colleague who's still learning the ropes.

So yes, AI can be a teammate. Not the loud one in meetings or the hero who saves the day, but the quiet contributor that has the potential to make everyone a bit better at their jobs.

And maybe, for now, that's all it needs to be.

The real question

The next era of collaboration is about how we work with each other, even when "each other" means code.