Where AI has its limits¶
With all that AI can do for project management and resource planning, the question arises: If AI detects problems in planning can it not simply solve them itself instead of "just" making suggestions? The sobering answer: not yet, unfortunately, despite all the developments currently taking place in the field of Artificial Intelligence.
Cascade effects and other disasters¶
An AI (or a simple algorithm) can very well calculate a constellation of work and people that solves the problem of overload. However, this can lead to virtually all projects having to be completely rescheduled, and that, since we are talking about computers here, quite possibly once per second.
Moving a work package with 5 allocated resources by one day can trigger a cascading effect that adjusts perhaps 50 other projects and hundreds of scheduled works. This may still be feasible in a project with an exclusive set of resources. At the latest, this is impossible in a cross-project planning, such as holiday planning.
An example:
The AI might consider it sensible to postpone employees' holidays by three months. This is impractical and would have a very negative impact on the acceptance of the AI. In addition, an AI often does not have all the information to find a solution. Projects consist of deadlines, costs and, above all, quality - in other words, content. An AI cannot decide whether a planned process can be realised just as well with slightly less amount of work and acceptable quality losses.
Human aspects and other unknowns¶
A good project management AI knows human behaviour, learns its patterns and can simulate them to a certain extent. However, human solutions to concrete problems are still unknown to the machine. Motivating employees and focusing them on the important tasks in order to achieve the goal has been difficult or impossible for software up to now.
Therefore, the project manager or the Scrum Team must come up with a solution themselves - but software can strongly support them. Whether this imagined solution can be implemented is subject to two primary, equally important aspects. On the one hand, the solution must be realistic in terms of content, especially if the amount of work of a job is reduced.
Secondly, the solution itself must be feasible, i.e. does it really solve the problem if we reduce the amount of work in a work package by 30 % or postpone the work package? An AI can also perform this evaluation task. The person using it simply enters the solution into the software. Since everything is calculated in real time, he sees immediately whether the problem has been solved but a new one has arisen. It is a collaboration between human and machine assessment, a procedure according to the principle of "trial and error“.
Real-time and other inaccuracies¶
When the software works in real time, it is a great help. The change is then simply "tried out" in the project plan. If the risk disappears, everything is fine and the solution can work. If not, an alternative or enhanced solution must be developed.
Real-time processing also means that the data available to the software must be reasonably up-to-date and realistic. Project managers have little or no opportunity to identify and solve problems if the team only enters the actual work progress into the computer at the end of the month. The simulation of an AI is then based on data that is not up-to-date. The results of the algorithms and the AI are therefore also not reliable.
Therefore, companies must attach great importance to the project staff updating the work progress and the amount of work daily if possible, but at least weekly. This is the only way that the project manager can anticipate future risks with the help of the software and take preventive measures.
Interestingly, the employees who are most overworked and stressed are also the ones who update their data the least. This is a logical interaction that leads to an endless loop of problems.
The same correlation can be observed at the management level. The projects of those project leaders who do not maintain their plans or simply ignore the risks that the machine uncovers according to the principle of hope also cause the most problems. Because they disrupt the other projects through their project mismanagement. However, it is not uncommon for these project leaders to complain about the challenges they have caused themselves.