Get Ready for an AI Revolution in Public Sector Project Management

Let's face it. Being a public sector project manager of large projects is no piece of cake. Among their challenges is that PMs must ensure their project's compliance with torrents of government policies, regulations and bureaucratic processes all while trying to keep delivery on track. In practical terms, the effort expended in compliance activities can devour a great deal of project management staff time that could otherwise be used to advance core project milestones and deliverables.

Compliance isn't necessarily difficult or creative work as much as it is methodical and time intensive. Yet time-pressed PMs can find themselves tapping their skilled project staff to address compliance tasks. Efforts expended on compliance can often displace work expended on core project deliverables, resulting in a vicious circle of schedule delays, increased costs, change requests and client dissatisfaction.

Fortunately, help is at hand for beleaguered project managers in the form of some fascinating advances in technology.

The rise of LLMs

Unless you've been living on in a cave for the last several years, you've likely heard of artificial intelligence and machine learning. But what may have escaped your attention until recently is a subset of AI called Large Language Models (LLMs). LLMs are seen by many as the next disruptive technology. In a nutshell, LLMs seek to give computers the ability to understand language as it is spoken or written. They parse vast quantities of text for its salient features and create relationships amongst these features in terms of the statistical strengths of their associations. Once the LLM is "trained" in this way, it can be prompted through request and questions to produce an output.

LLMs trained with very large numbers of features can give the appearance of having expert-level experience and reasoning capabilities across many knowledge domains, including project management. To illustrate, let's take a look at a couple of examples from an LLM called ChatGPT created by Open.AI.

I provided ChatGPT with this prompt:

Me: I'm a junior project manager. My boss said that I must pay particular attention to one of the sides of the "iron triangle" if anyone is to get any time off at Christmas. What was he referring to?

ChatGPT: Your boss was likely referring to the time aspect of the "iron triangle" of project management, also known as the triple constraint. The "iron triangle" refers to the three interdependent constraints of a project, which are time, cost, and scope.

If your boss wants you to pay particular attention to one side of the triangle, he may be suggesting that meeting the time constraints, such as completing the project in time for the Christmas holiday, is of utmost importance. This might mean that you need to make trade-offs with the cost or scope of the project in order to meet the time deadline.

In general, adjusting one constraint of the triangle often affects the other two, so it's important to keep a balance between all three in order to ensure the success of a project.

Wow, impressive! A few things jump out in this response. Firstly, ChatGPT correctly associated the "iron triangle" colloquialism with triple constraint, which is the official term for the concept. Secondly, it correctly determined "anyone" implies the people working on the project and that the reference to time off at Christmas is related to the schedule component of triple constraint. Finally, it helpfully provided the potential implications of adhering to the boss's prompting.

OK, let's assume that I learn concepts better when in verse.

Me: Write a haiku to help me learn more about this.

ChatGPT:

Iron triangle rules the game,

Cost, time, and scope must be kept,

Balanced, success is claimed.

Well, no poetry awards for this one, but it demonstrates ChatGPT's ability to combine compositional rules with a subject matter knowledge base to produce the required output. It also demonstrates contextual persistence. ChatGPT knew that I was referring to the concept that was being discussed in the previous prompt so it shows an ability to retain the context of interactions.

AI Help for the Public Sector Project Manager

So let's have a look at how these capabilities can help reduce compliance burden on PMs and reduce project review process time.

Let's assume that we have an LLM that's already extensively trained on project management knowledge. Now extend that training with all governmental policies, directives, and standards relevant to public sector projects. Also supply it with the records of past government projects including their lessons learned.

What we have now is the beginnings of an expert assistant to the PM. When this tuned LLM is supplied with draft project documents and told what to look for, the LLM can instantly check for compliance to all relevant constraints, even those the PM was not aware were required, and flag these for project staff to review and correct before even leaving project office. This can vastly accelerate the current manual review process where service lines review project documents, each service line performing its specialized checks before returning the documents to the project office for edits. But shorter review cycles are just the beginning.

Let’s go further and give the LLM information on the project’s business context and details of the project's sponsoring and governing organizations including mission, structure, available resources, etc. The LLM is now is now equipped to assist the PM with detailed project activities like planning, project control, reporting and governance. The PM and the LLM can now work together to rapidly compose project documentation including charters, schedules, financial budgets. And because of the model's knowledge base, each document will be standards compliant as soon as they are created.

LLM sophistication is increasing at breakneck pace and will become increasingly capable of relieving the compliance burden of PMs. But as I stated earlier, this is just the beginning. How is your department positioning itself to begin this journey?

Originally published on LinkedIn, February 8, 2023.