Natural Language Processing (NLP) is a subset of machine learning that aims to teach a machine how to recognise and understand language and text, in this case evaluating project task commentary submitted via email and automatically assigning a RAG status. It has been used in a variety of domains, where NLP serves as a link between artificial intelligence and humans, integrating languages, computer science and automation. As the field of Deep Learning NLP has advanced, it is now appropriate for use in project management reporting, making it a valuable tool for process improvement immune to cognitive biases and misclassification. This article examines how this technology can be used to automate project risk classification by employing a trained DL NLP model to review, comprehend, and assign risk classification based on commentary submitted to the project manager or project management office in conjunction with office applications.

Machine learning, as opposed to intelligence programming, is the process by which a program learns on its own with the help of data and statistics. As more data is fed into the model, it learns to implement the most logical outcome. Large data sets, years of successful and unsuccessful project scenarios, such as inputs and outputs relating to risk, can be analysed by ML models to learn which input scenario will result in which output with the greatest degree of certainty.

Evolution of RISK classification

An industrial automated control system includes a human–machine interface. Consider programmable logic in automated industrial systems that use a defined sequential logic as an example. If A and B occur, C is triggered, and conditions are realised based on predetermined, proven factors. The system is optimised to produce the desired output in the most efficient way possible. The human machine interface is only used when intervention and control are required by design. We can now use the same logical framework based on textual commentary NLP models, instead of digital and binary conditions or mechanical gates we can utilise contextual data. Deep learning NLP models can be trained on large data sets containing years of project RISK data. Not only can the NLP models understand and associate textual inputs with outputs, but they can also understand context to analyse and classify textual scenarios based on historical knowledge.

In this example, the Project Manager serves as the Human Machine Interface, with NLP serving as the automated process for project risk classification reporting and escalation. As in the previous example, the HMI/PM should not be required to intervene unless required as an output of logic, such as a RISK classification or an unacceptable threshold of uncertainty, which would then trigger a subsequent sequence of controlling events forcing intervention.

The slow data transfer from human to machine, to human, and back to machine is the bottleneck in the preceding example. The risk of cognitive bias and misclassification in data transfer is high. By automating various processes, the Project Manager or PMO will be able to shift away from cognitive reasoning, which is susceptible to human error, within specific areas of the project management framework and instead function as an overarching, strategic controlling mind, essentially an optimised project manager.

The future

As we move towards the “Fourth Industrial Revolution,” Artificial Intelligence can be used more extensively at a basic level to optimise project risk classification and mitigation, resource scheduling and utilisation analysis, project performance reporting with advanced projections, and KPI analysis.

From the ground up, we investigated the fundamental implementations of a single subset of Artificial Intelligence that is process driven. The array of possibilities presented by this fundamental deep learning concept should have opened your eyes to the change transformation that is on the horizon.

Taking a more strategic and scalable approach, let us now investigate how Artificial Intelligence can have a strategic impact on the project management industry from the top down. How will it affect the industry from an organisational standpoint, and will we establish new methodologies, processes, and frameworks as a result? Will AI create a unified toolbox of tools and utilities to assist a project manager or PMO? The impact of AI in project management will be truly disruptive on a large scale across all industries, fundamentally redefining project management practice.

Risk analysis, mitigation, and forecasting are generally practices that accumulate as a project progresses, with the risk register being somewhat predefined to a minimalist level and updated reactively. RISK is inherently uncertain, and without the ability to predict the future with 100 percent accuracy, this will remain the case.

Nobody can guarantee project completion on time, within budget, within scope, and with agreed-upon quality; it is not yet possible. Risk, on the other hand, can be modeled on an endless application of historic data, machines can learn to calculate and forecast risk with incredible accuracy more so than any human counterpart. Building this function into the project management industry as standard will be a core evolutionary step. Learning is essential; however, the industry is failing to learn from previous projects and processes, adopting what worked and avoiding steps to failure.

This process must be implemented on a regulatory level so that Artificial Intelligence can model all industry data to provide a standardised risk, management, and control index that learns and evolves, allowing for better project management practice that is automated, controlled, optimised, and risk averse. The next evolutionary step is for Artificial Intelligence to be used to address critical industry flaws; projects should be integrated not independent.