With 93.23% Accuracy, AI Can Now Measure the How Good a Software Project Manager You Are

If you are a software project manager, watch out. Artificial Intelligence (AI) could complete your next performance review.

Researchers in Portugal are currently trialling an AI system capable of predicting the competencies of software project managers with 93.23% accuracy.

Project managers capable of solving complex problems are essential in every software team. Which means this research could have huge implications in evaluating the success of software project managers.

We may also start to see this system in the recruitment of new project managers.Companies would be able to quickly check the suitability of candidates on a quantitative set of criteria.

Currently, the AI system works by measuring 13 sets of criteria. This includes the ability of a project manager in maintaining team communications, making the team goal orientated and their effectiveness in running team meetings to name a few.

To measure these qualities the system uses Artificial Neural Networks (ANN). ANNs work in a similar way to our brains, learning from examples and past experiences. Therefore, ANNs are most useful when it can be ‘taught’ how to correctly complete the task.

This is true of the AI system currently being researched in Portugal. It has been trained using historical data from previous projects and teams. The predictions it makes are then compared with accepted competency models. In which it currently has an accuracy of 93.23%.

But the capabilities of this AI system are somewhat limited in its current state. As a result, it may not give a true representation of the software project manager in question.

Firstly, it is limited to testing only select number competencies that software project managers need to have. It is currently unable to manage the large data sets required to make such decisions.

Additionally, the accuracy of competency estimations is dependent on the historical data it uses. Like humans, the accuracy of the system decreases when placed in unknown situations. Meaning the correctness of the system is dependent on the amount of historical data which can be collected and stored.

Needless to say, this research is only in the early phases of testing. Expect to see the accuracy of competency predictions of software project managers using AI to increase in the future.