AI for Major Projects
In 2025, how would AI change project management?
Imagine an assistant which has lived through 20,000 projects; and which can actually identify upcoming problems and alert the project team before problems become too complex and costly to solve. How would that work? The machine, compared to humans, is extremely adept at comparing terabytes of data in a matter of seconds. The machine would scan every terabytes of writing exchanged on a project: it would identify inconsistencies, suggest previous similar situations and connect the dots to predict. Some may be surprised that I am describing a system available in
2050 rather than in 2025; yet I deeply believe this is possible and will profoundly reshape our project management ecosystem? That’s why it is vital to be ready.
The current situation in 2020
Today, we hear about artificial intelligence (AI) everywhere (in fashion magazines and even to better brush one’s teeth); and yet, in large projects
(nuclear powerplants, bridges, roads, etc.), projects continue to run with scattered word, excel, pdf, emails, etc. without any assistance from the
machine. Lack of assistance means that between the beginning and the end of a project, the system does not acquire any intelligence that could be
used to help the project team later AI will less and less be optional in coming years, because:
1/ there is more and more communication, and it is becoming humanly not manageable. As the data is more and more scattered, we need the help of
the machine to do a preliminary sorting in order to be able to digest this data and take the right decisions.
2/ the financial stakes of the delay have a significant impact on company margins: today average cost increase of large scale projects is 80% of original value and the average slippage is 20 months behind original schedule (Mc Kinsey1). Average claim is 33 M€, average length of dispute is 17 months (Arcadis report2). We need machines to monitor blind angles and help reducing the exponential financial consequences of delays as soon as possible by spotting weak signals in the documentation.
3/ the most competent members of the project team are retiring with their expertise, how will the younger generation continue to carry out ambitious projects without having access to all this experience? It is essential to find digital ways to make this experience available and to process the
amount of data we create during the course of a project in order to continue to deliver.
We need to plan to be where we need in 2025
The belief that AI is magic and will instantly solve all problems, depending on the number of millions euros invested, is absolute nonsense. Deploying an AI system is a complete business roadmap that can take from two to five years, depending on how quickly a company can modify its processes and
- Put strategies in place to ensure that the right data is collected for future use The first step should really be to map all the problems you want to solve using AI, and analyze the gap between existing data and
- List all the difficulties to deploy AI Multiple difficulties will emerge while deploying AI for instance: data centralization, fear of big brother, lack of clarity, etc. Those difficulties need to be listed as soon as possible so they can be tackled one by one as soon as possible.
- Give a data coloration to project teams. There is a need to train the current professionals to data, new jobs are expected:
- AI trainers: project specialists who will create the initial labelled datasets to bootstrap the AI system
- Project Analysts: project specialists who will be able to verify the output of a machine and suggest action plans.
- Project AI Strategists: project specialists who will create an AI roadmap to ensure that data collected are the one required to
answer to the right AI questions.
Those who will gather both the expertise of project management with a color of Artificial Intelligence will become the most wanted professionals.
Among the opportunities brought by an AI
Today‘s project teams rely primarily on individual memory, knowledge and assertiveness to successfully complete projects. Tomorrow, project teams will be assisted by an assistant who has been trained on thousands and thousands of projects, and who will be able to suggest anomalies, problems and opportunities thanks to its unique ability to compare years of data simultaneously. And because fact-checking is effortless, you can expect the facts to beat „getting the story right“.
A dynamic use of the project documentation
At Lili.ai, we are turning the project documentation into a dynamic system of beacons. For instance, we turn existing excel risk registers into a set of pre-trained classifiers scrolling through the data 24/7 to identify actively new upcoming risks, known patterns, etc. Example: we are able to identify peaks of mentions of« difficulty» related to the activity of “tunneling”. This is an early warning that someone should have a chat with the tunneling teams. Similarly, we have successfully extracted
a Gantt chart directly from the mentions of activities in the emails. The said Gantt chart is profoundly different than the theoric alone; as
activities are not at all happening during the beautiful unified bar but throughout a longer period.
A holistic view and analysis of the project
The data is used to create a centralized knowledge graph offering a unique point of entry based on semantic themes. The knowledge graph enables a uniquely fast and exhaustive search of all topics related to a specific topic. Hence, knowing that a delay on an activity will actually impact another one; all this in a dynamic manner.
AI for project management is certainly a natural step given the amount of data to be processed, but there is no doubt that this transition will not be easy and will require some planning. The main promises of artificial intelligence are better, cheaper and faster capitalization. Tomorrow, each project will be equipped with a digital project assistant that will make information more fluid and give people more time to perform human tasks rather than mechanical tasks (copy-paste, compare, track, etc.).
With a unique real-time data processing capacity, what kind of ambitious projects will be possible?
To read the full magazine, please download it at https://pmi-german-chapters.de/