Ronald RichardsFounder and CEO (PMP)| SCIVE Labs, LLCGreenlawn, Ny, United States
When planning that next project and determining whether or not it is worth doing, how often do we carefully review Lessons Learned? All that extra work! Perhaps not so if AI is employed.
I think that it may prove itself indispensable in this respect. Saving Changes...
I look at the application of AI to PM lessons learned as on par with credit card companies using spending patterns to detect fraudulent activity. The ability for AI to work with freeform text makes it ideal for using documented lessons learned or problem reports to find prior patterns that apply to your current or perhaps future situation. Saving Changes...
Markus KopkoAI Enabler for Project & Program Mgmt | Founder PMotion.ai / The PM
AI Coach| PMotion.aiHamburg, Hamburg, Germany
Dear Ronald,
Employing AI in reviewing lessons learned can greatly enhance the planning process for new projects. It makes the examination of past experiences more thorough, efficient, and insightful, aiding in informed decision-making. As AI technology continues to advance, its role in project management, particularly in knowledge management and predictive analytics, is likely to become increasingly significant, turning what was once a daunting task into a streamlined, integral part of project planning.
BR,
Markus Saving Changes...
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
Remember: AI will take the available data. In the last time I worked in an initiative to create a pluging to one generative AI tool that worked with our data inside our knowledge management system so, in that way, we can consult lessons learned and lot of other things. For example it helps to create proposals to clients. Saving Changes...
Digging through reams of lessons is just one of the problems with the whole concept of "lessons learned". While it can be partially addressed with use of AI, issues such as a lack of specificity or low quality in lessons, a lack of context, and staleness won't be addressed by AI.
There are better ways to increase organizational learning which should be considered including communities of practice and baking what we learn directly into our standards and practices.
Learning & Innovation Research Manager| Project Management Institute (PMI)Spain
Hi Ronald, I have read an interesting article by Sushant Prabhu on LinkedIn (see link below), describing the data path in building an ML Model for E-Commerce Fraud detection and I was thinking about how to apply this path when creating a lessons-learned AI lifecycle in our companies. I would see it in this way:
- Problem statement and objective. Our objective is to create an AI lessons-learned system to detect risky projects and prevent losses in our company.
- Dataset. We will decide the sources of our data, and where the system will be learning from. For example, if we have project data stored this could be a great source, if not we could buy synthetic data, and see what other companies in the sector are doing…
- Feature engineering. We can extract features that we consider relevant to our projects. As project managers here is our strength These could be, for example, geography-related data, data about our teams, if they were dedicated, time zones, causes of delay, estimation methods, planning approach, management method...
- Data preprocessing. We would need the help of our data scientists or tech people to clean data and ensure consistency so that the obtained data is relevant.
- Exploratory data analysis. We would explore the data obtained and see if we can identify key indicators and understand patterns, for example, if we see that projects in an X zone are severely delayed when starting in Winter, or if dedicated teams perform better than shared ones, or if our agile approaches are delivering better results than hybrid ones.
- Data splitting. Data is divided into training and testing sets to evaluate model performance.
- Modeling. We choose the right machine learning model so that we can identify and classify specific lessons learned leading to successful or failed projects.
- Hyperparameter tuning. Model parameters are fine-tuned to optimize accuracy and minimize incorrect predictions.
- Results (output). The lessons-learned system to detect successful and failed projects is now in place and will help us reduce the chances of embarking on risky projects.
- Feedback for improvements. Continuous monitoring will be essential to keep on improving the model´s accuracy.
How do you see it? Is it interesting to know where project managers have our strengths in this data-driven world?
PMI Team Member
