Harnessing AI to Mitigate Biases in Project Management

Introduction

During the holidays, one of our longtime family friends, who also happened to be a programme manager at one of the "magnificent seven," visited our home with his wife and two wonderful daughters. We both have been in the project management profession for quite some time now. Despite being surrounded by four kids between 4 months and 8 years old, AI entered our conversation silently. The impact of AI on project management as a profession and the technology industry as a whole kicked off the discussion. His original thought was that project managers would always be needed as long as there were politics in the organisations and projects. We reassured ourselves that our jobs were safe for the time being, laughed for a bit, and continued enjoying each other's company. However, this discussion got me thinking about the broader effects AI would have on the field of project management and the behavioural aspects of individuals involved in any project, which is the key constituent of office politics. 

Behavioural Bias in Project Management and Decision-Making 

Have you ever heard one or more similar statements during your projects? 

"Once this product goes live in production, I am 100 percent sure that we will have a large number of users who would subscribe to it at a premium price." (Overconfidence bias) 

"RFP is a waste of time and resources. I already know a vendor who provides the best technology to solve all our problems." (Availability Bias) 

"Why change this process or technology when it is already working." (Status quo bias) 

"We need to spend all the allocated budget for this year so we can get the same or more budget next year." (Illusion of Control) 

"There is something wrong with the project plan. In my previous organisation, I completed a similar project in half the duration." (Conservatism Bias) 

"Project is delayed because the business did not provide the correct requirements." (Hindsight Bias) 

"Who will be responsible if something goes wrong? I do not want to hear about any changes in the plan." (Loss aversion bias) 

"I have been in this industry for a long time. Agile methodology is always better than waterfall (or vice-versa)." 

"All the bugs are critical and must be resolved on Day 1." (Framing bias) 

All these statements reflect different types of biases. One will observe that these statements come from people generally in higher organisational positions. It is not impossible, but it is also not easy for project managers to point out these biases and plan projects effectively. A lot of effort and collaboration goes into the process of coming up with a project plan. These biases can skew decision-making and derail the planning process, leading to suboptimal project performance. A good project manager could develop a great project plan, but only a great project manager could defend that plan from all these biases and not succumb to external pressure. 

Whether working on multiple tasks within a project or working on various projects simultaneously, most of us multi-task regularly. Cognitively busy individuals are more likely to make biased choices. Also, most decision-makers are overconfident and rely on intuition more than they should, as cognitive effort is mildly unpleasant. To worsen the situation, decision-makers could even go to the extent of not accepting the new information that would not support their original decision. Even project managers are equally guilty of this as we make various decisions while managing our projects. 

However, could the project manager identify all project biases and recall all the lessons acquired from previous projects? Is it possible for the project team to make fewer decisions? Can a project team maintain constant vigilance over one another's behaviour without betraying trust? 

How are Project Managers Handling these Challenges Today?

Soft skills—One single most important requirement for a project manager is to have soft skills. Project success depends heavily on how the project manager communicates and manages expectations with stakeholders, how they build trust and alliances within and outside the team, and their ability to influence senior leadership's decisions to solve real-world problems. 

Project management skills—Whether managing RAID logs, performing a cost-benefit analysis, or deepening understanding of project management methodologies, a project manager who has acquired project management skills through experience or education is better equipped to tackle these challenges. 

Another critical factor for the project manager's success in driving the project's success is control of the project budget and the organisational structure. Project managers in strong matrix organisations in full-time roles with high authority and control of the budget are empowered to handle these challenges compared to project managers in weak matrix organisations with little to no authority. Most Project Management Offices (PMOs) work under a strong matrix organisation structure. 

How will AI Change Project Management?

Chatbots and generative AI always take centre stage in discussions about artificial intelligence, and project management is no exception. Artificial Intelligence (AI) is a game-changer that could assist project managers in becoming more productive in their daily work by automating reports, scheduling activities, summarising meeting minutes, etc. However, is that sufficient to fully employ AI? Not. AI can further assist project managers and answer some of the questions raised in the previous discussion. 

According to a study on entrepreneurial decision-making, startups that use a scientific method to test and challenge their beliefs have more tremendous success than their counterpart who rely more on heuristics. However, we have all witnessed leaders make decisions based on their gut feeling, familiarity with the recommendations or those making them, and how the information is presented. AI will improve consistency in the decision-making process by focusing on relevant data and supporting evidence rather than bias and prejudice. Using decisions that have already been made in comparable circumstances in the same or other projects may also help to minimise the number of decisions that need to be made during the project. This will lessen the cognitive strain that decision-makers must endure during project decision-making and free up their mind for other critical decisions. 

How many project managers diligently record the lessons they have learned, no matter how big or small, and then use those lessons in their subsequent projects? Most project managers use what they've learned from their prior projects in their new ones, although it mostly depends on how much they can remember. All of the lessons learned from the last project or from initiatives we completed years ago are tough to recall. Additionally, reviewing the lessons learned from previous projects is a cognitively taxing and practically impossible task for project managers. AI could help project managers not only document these lessons learned consistently by analysing our meeting transcripts and other project artefacts but also make recommendations to apply these lessons in the next project in appropriate situations. 

Imagine AI studying new regulations, evaluating them against an organisation's current policies and procedures, and recommending changes. Helping organisations collaborate on how to address new regulations collaboratively. Shortlisting vendors based on requirements, responses, and cost-benefit analysis. Project budgets and schedules will be shortened as a result. 

AI can analyse enormous volumes of data, forecast possible risks, and advise project managers, teams, and other stakeholders on reducing these risks. Prioritising features and bugs would be done with statistics, not gut feeling. Machine learning algorithms and predictive analytics can spot trends that people might overlook. AI will simplify the process of performing root-cause analyses, evaluating results against benchmarks, and validating the initial idea. Although AI may not ensure project success, it will assist the project team in managing their expectations by forecasting outcomes early in the project. 

Artificial intelligence will be able to detect project biases and offer practical solutions by combining qualitative and quantitative data. Additionally, team members and leaders would be more willing to change their behaviour based on these AI insights rather than someone else in the project pointing it out. AI could help avoid the battle of egos. 

Challenges

So why not just wave the magic wand of AI and solve all the problems of project managers? Not so fast. Artificial Intelligence relies on large amounts of data for its efficacy. LLM (Large Language Model) is trained on a large amount of data to make it work effectively. The biggest challenge will be gathering and organising data for deep learning from a project management perspective. 

Some organisations in the project management industry are doing a great job of bringing project managers together, but the project management industry at large is highly disconnected. Organisations do not share data on projects with one another due to confidentiality reasons. Even within the firm, there could be an information barrier between project managers. Unsuccessful projects, the best learning source, are rarely discussed constructively. Adopting AI tools could be expensive, and PMOs always run on a tight budget. 

Internal Data Collection

Firms already use many project management tools, which project managers use to collect this data manually for reporting and other purposes. We will have to start applying AI tools extensively to collect data automatically from projects in a structured manner within a firm. A large Language Model (LLM) could then be trained on this data to prompt past learnings, predict risks, and recommend decisions. 

External Data Collection

Conventional methods like surveys and interviews with project managers and research papers from academics will be of great importance in creating a knowledge base. Retrieval Augmented Generation (RAG) could enhance the accuracy and reliability of generative AI models with facts fetched from external sources. A framework would be needed to train LLM on project data from external sources. The data-gathering exercise will involve collaboration between project managers, firms, and AI SaaS providers. Inspiration could be drawn from open-source platforms like GitHub, Stack Overflow, etc., where developers share and find solutions to specific programming problems. There is a high chance that a project being done by one organisation has already been done by another. By sharing project knowledge, organisations can collaborate and create a knowledge base. AI tool companies could also encourage firms to share project data by incentivising them with reduced costs for AI tools. 

While project managers will be learning AI, AI will also be learning with them. There is going to be a lot of handholding needed to make AI successful in project management and help project managers and organisations succeed. Initially, AI will be more successful in specific projects than others, but it will evolve continuously, and stakeholders will build trust in it over time. With patience and a cautiously optimistic mindset, project management leaders in supportive organisations will set new practices and standards by bringing AI into project management. 

Conclusion

Project managers will remain in demand, but their working environment has already started evolving. Project management will continue to be a combination of art and science, but AI will bring more science into the equation. 

AI will help reduce the number of decisions that decision-makers have to make and make them self-aware of their own biases while making decisions. It will increase the probability of success for projects and organisations overall. 

But there is a long way ahead to collect project data, which AI could use to train itself to help project teams. 

New automated methods would have to be applied to collect this data from projects. Project Managers, Organisations, and AI SaaS providers will have to collaborate to bring the full potential of AI to project management. 

Reference Literature: 

1. Process Groups: A Practice Guide, Project Management Institute, 2023, p. 46  

2. Thinking, Fast and Slow, Daniel Kahneman  

3. Atlas: Few-shot Learning with Retrieval Augmented Language Models

4. Arnaldo Camuffo, Alessandro Cordova, Alfonso Gambardella, Chiara Spina (2019) A Scientific Approach to Entrepreneurial Decision Making: Evidence from a Randomized Control Trial.