I was a project manager  -  given the chance, here's how I would have used AI - Part 2

Following on from Dev’s recent article, this is how I would use AI as a Project Manager if I were to enter the PM arena again.

When I reflect on my experience as a project manager, the defining memories are those unexpected challenges that blind-sided me and the wider project. Over 7 years, this included:

🚱 Extensive, unexpected contamination

🔥 A major fire destroying a large part of the site

🪧 The need to accelerate to meet immovable deadlines 

At the time, these events felt shocking and distressing. Could someone be seriously injured or killed by the fire? How will this impact the success of the project and the project stakeholders? Will I get fired if there is a cost and time blow out?

With time, I have come to ask a different set of questions. Was there a way to be better prepared for these eventualities? Could these risks have been identified before they occurred? Could we have adapted better and faster to the challenges as they emerged?

I was recently introduced to an interesting concept known as Emergence that explains different types of risks, our ability to predict them, and the implications for project managers in complex projects.

A quick explainer on Emergence

Emergence is the idea that projects are a complex and interdependent system, and their outcomes are impacted both by the component parts of the project as well as the outside environment. This makes those outcomes very difficult or impossible to predict.

While the risks that materialised on my projects may not have been obvious or expected, they were discoverable with access to the right information and expertise, i.e. they were from Zone 2.

“[Zone 2] is the zone of blind spots where projects are exposed to unknown tangible, measurable or otherwise well understood uncertainties because of gaps in knowledge or research. The challenge here is that we try to grasp the unknown future based on the model of our experience in the past and ‘because of cognitive limitations’, it is neither possible nor desirable to represent every detail that constituted that experience.” 

How I would use AI as a PM

Avoiding these catastrophic impacts, hidden because of the lack of access to information or expertise, would be a critical problem that I would try to tackle using AI as a PM if I got back into the ring.

So how might I have applied tools that exist today, or that could be built using AI, to tackle the specific challenges we faced?

Fire 

The site fire was a low likelihood, high impact event that might have been identifiable if we were able to uncover the clues in the environment before the fire started. Image and video based site monitoring, with managers and workers wearing hard-hat mounted cameras to capture progress, is increasingly common on projects. This data could be used to identify and alert teams to fire risks using AI models trained on images known to represent elevated fire hazards. 

Site monitoring products available today include OpenSpace, Oculu, and Doxel, though they don’t yet have specific fire risk identification features. 

Contamination

Core and test pit samples were collected in preliminary investigations on the site, but this was insufficient to determine the extent of contamination. An AI model could be used to better predict the likelihood and extent of contamination with the extensive soil sample data that exists today. The mitigations from this discoverable risk could have been commercial (allow time and cost), contractual (exclude from scope), construction sequence based (staged decontamination). 

Beyond the specific issue of contamination, this highlights the impact of tail events and the importance of modelling them accurately, and knowing that they won't always happen on your project but they have happened somewhere, and knowing that potential impact lets you plan instead of react.

I’m not familiar with any tools that solve the problem of contamination in this way, so perhaps this is an opportunity for an aspiring entrepreneur!

Acceleration

The combination of fire, contamination, and other outlier events resulted in the need to accelerate the project to meet fixed deadlines. One of the tactics used to solve this problem was to work 7 days a week, rather than the 5 days a week that was planned. This was costly from a financial and morale perspective in a project that spanned several years.

Using a tool like nPlan, had it existed at the time, would have provided a systematic way to review the most impactful mitigation strategies that could heave delivered the same results at a lower cost. 

Conclusion

Emergence makes projects complex and difficult for PMs to predict, but AI can be the tool to uncover the information that is hidden from view or siloed within the brains of experts who may not always be available.

All of these tools would allow me as a PM to intervene in risky situations and avoid their impacts, and successfully deliver projects more often.

What about you?

How do or would you use AI as a PM?