I ran an experiment to see whether AI could manage simple projects in the real world. I granted it the authority to act as the project manager for a personal project that had stalled, and I followed it’s instructions.

The video above shows the end to end experience, but I made some key observations from the experiment around the tools strengths and weaknesses as a project manager.

• Indecisiveness – I granted the tool the authority to make decisions, but it never really managed to make a decision. I would sometimes offer a recommendation, but never truly embraced the role and the authority.
• Efficient – the AI was incredibly quick at producing useful project documentation like the charter, plan and risk register. It never got it perfectly right first time – as would be the case with a human project manager – but it adapted and updated incredibly fast.
• Passive – unfortunately the AI is not currently capable of chasing me for updates, following up, contacting people to remove blockers. It is dormant until you engage with it.

Whilst the experiment revealed ChatGPT is not yet ready to lead projects yet, the future is bright for AI. Chatbot sophistication is improving all the time, and automated AI will likely make a huge impact when it arrives (removing the passive issue I mentioned earlier).

Check out the full video if you haven’t already, and let me know your thoughts on potential future experiments to test AI capabilities.