Project Management Challenges with Artificial Intelligence

In the past five years, there have been numerous articles discussing how Artificial Intelligence (AI), can and will benefit the field of project management. As technology increases within the next two decades, AI is expected to replace humans in many of the simple and mundane tasks that are part of project and program management activities (Grace et al., 2018). The most common applications of AI are expected to be part of estimating and controlling project cost and time, as well as resource management by determining the strength of employee qualifications from assignments to selected project activities. Other applications of AI are expected to included risk reduction practices, improved monitoring and control, status reporting, identification of anomalies, and correlations between projects (Ong and Uddin, 2022).

Applications of AI already exist in today’s businesses and industries. Examples of AI use include Alexa, Siri, Amazon’s product recommendations, dating apps and fitness trackers. (Marr, 2020, and Dalcher, 2022).

The growth of AI can be attributed to several factors, especially:

As the growth continues, disruptions are expected in the business models that companies use. The project management landscape will change significantly. Companies that are unwilling or unable to adapt to the changes generated by AI practices may not survive.

Recognizing the need for AI applications has been discussed in numerous journal articles. Many of the articles focus on the benefits expected and areas of applicability but spend little time discussing the challenges that companies will face. The intent of this paper is to discuss the issues and challenges that may prevent successful AI implementation, at least in the short term.

Customer Reaction
Gaining support for the implementation of AI practices should be significantly easier when the deliverables and outcomes of projects are for internal customers. External customers may not be supportive of the influence and decisions that AI software might have on their projects.

During competitive bidding practices, contractors may be required to include in their proposals how they plan on using AI during the execution of the customer’s projects. Customers are interested in not only the final deliverables, but the methodology used to achieve the results. Some customers may be hesitant to allow AI to make certain decisions on their projects. In the short term, contractors may find it necessary to have two forms of project management: traditional project management practices and AI project management practices. Customers may be given a choice of which project management approach they prefer.

In the past, customers were often prevented from participating in many of the decisions made by the contractors. As customers became more knowledgeable in project management, customer involvement was seen as beneficial and welcomed. Some customers may be unhappy that their involvement in projects is now being replaced by AI practices. Contractors using AI practices may find it necessary to educate customers on how AI will impact their projects.

Project Management Maturity
Project management maturity is usually defined in terms of continuous improvement efforts in the forms, guidelines, processes, templates, and checklists used in project management. About two decades ago, organizations and professional societies such as PMI^® created project management maturity models (PMMMs) to assist organizations in their quest for project management maturity.

Most maturity models were based upon questionnaires or assessment instruments related to how well the company was aligned to the processes, knowledge areas, and domain areas identified in PMI^®’s PMBOK^® Guide. The PMMMs will now have to go through continuous revisions to allow for the impact of AI practices on project management maturity. Not all knowledge areas are expected to undergo AI transformations at the same time. The easiest knowledge areas for including AI practices might be time, cost, and integration management. Other knowledge areas may be more challenging in accepting AI practices even if they are connected to areas that are heavy users of AI techniques.

Educating Project Managers
There are numerous university and private sector companies offering training programs for projects managers. Training normally centers around the knowledge and domain areas in the PMBOK^® Guide and the Standard for Project Management. Unfortunately, as of now, concepts related to AI are not included.As technology replaces the simpler or mundane project and program tasks, project managers can be expected to be integrated into the higher levels of project management such as in fuzzy front-end project selection and prioritization processes, strategic planning, and strategic business decision-making as it relates to projects.

For an understanding of AI concepts and how they are used in projects, PMs may require training that include tools related to statistics, software development, advanced data analyses, trend analyses, and data intuition analyses. The tools are expected to be different for each life cycle stage of a project.

The educational modules related to AI will be based upon information and data collected from a variety of internal and external sources but intended for use in the parent company. As such, the use of trainers external to the company for AI education may be restricted to general AI knowledge training. There may be many customized company software tools and because the data may include company sensitive information, how the tools should be used based upon the information in the knowledge repository may require internal training programs possibly conducted by one of the parent company’s project management offices (PMOs).

Educating Team Members
For almost five decades, project management practices have used documents such as the LRC (Linear Responsibility Chart) and the RACI Chart (responsibility, accountability, consulted and informed) to assign responsibilities to team members. The charts were filled in at the onset of the project and updated as necessary as the project progresses.

By retaining the information in the charts, companies can use AI practices to assign the best qualified resources to project activities based upon historical data. The goal is to find the right skills for the right job and to identify a resources shortage if one exists. Additional information other than skills that may be collected might include employee historical performance review data, education, training programs attended, employees/stakeholders they have worked with in the past, and types of project assignments.

While this sounds like a good idea, team members may not be as open-minded as project managers when it comes to accepting AI as part of their job. Two of the concerns that employees will face include: (1) people may be afraid that AI will take away jobs, and (2) people fear failure (Wang, 2019). Employees believe that certain assignments may provide them with more opportunities for career advancement and may resist assignments made by AI. Employees may not be given a choice of assignments as they had been given in the past. Another concern is that AI may end up making decisions that employees made in the past and their jobs may no longer be required. Finally, AI may make decisions that the employees might not have made. If the decision turns out to have an unfavorable impact on the project, the employee may see this as a reflection on his/her ability.

Employees will be required to use AI technologies as part of their job. This will need to be integrated into all training courses to gain support and effective use of the AI software and data.

Risk Management
Project teams have great expectations that AI practices will help them reduce project risks by analyzing past data and running multiple scenarios to determine realistic outcomes and the lowest risk path. Unfortunately, there are several challenges that must be overcome before this becomes a reality. All projects today seem to be impacted by the VUCA (volatility, uncertainty, complexity, and ambiguity) environment such that many estimates and forecasts are at best intuition. Reducing uncertainty may be achievable but not a complete elimination of uncertain events.

Some degree of “humanness” must exist in all risk analyses activities. Even if an abundance of data exists in the information warehouse, the necessity for involvement and understanding of the decision-making process by humans will be necessary.
Projects are managed by people, not tools. Regardless of how much technology will be included in project management, the human touch of collaboration and interaction among team members and stakeholders will not disappear. As stated by Gupta (2020):

We cannot discount the human touch aspect when we are dealing with humans. Humans are social animals. We want to work with each other [rather than just with machines].

Ethics
AI is expected to provide the project management community of practice with more possible outcomes and the opportunity for better decision-making. The complex analytics track performance, look for trends, and make assumptions about where the project is heading. This assumes that the data inputted into the system was not prone to human error and that the outcomes were calculated in an ethical manner. As stated by Dalcher (2022):
Humans struggle to acknowledge and communicate tacit knowledge. While machines are more capable at capturing the aspects that we cannot articulate by observing and deducing patterns, we are often at a loss to figure out how a machine system has reached decisions.

AI systems have difficulty understanding biases and the recommendations made because of the biases. Brynjolfsson & McAfee (2017) discuss hidden biases that can exist in the data, and the difficulty in understanding and correcting errors.

Data gathering and data analytics are always prone to human error. AI cannot tell if critical data is missing. On projects related to R&D, innovation, and new product development, limited information may exist for reliable AI predictions. Also, there may exist multiple definitions of success that could lead AI analytics in the wrong direction. As an example, a company may be willing to invest significant funds into a portfolio of projects designed to operate at a loss initially to penetrate a new market. AI may be unfamiliar with the scope and misinterpret the real definition of success. This could result in faulty conclusions from AI software. AI is prone to errors especially if the database contains corrupt information. Legally, AI software cannot be held accountable if the wrong decisions are made and leads to bad consequences.

Another issue with AI ethics relates to the security of the information in the database. Security standards must be established on how the secured information is used and who may have access to the information without violating the right to privacy.

AI Penetration
AI penetration is expected to be part of all project management practices over the next 20–30 years and will occur slowly because of the potential for significant resistance. PMOs will most likely take the lead and help introduce AI practices in the project management areas of knowledge that might provide the least degrees of resistance. Later, as company-wide training programs are launched focusing on the use of AI practices, acceptance should be a bit easier once employees recognize the impact that AI can have on decision-making, reporting, forecasting, predictive analysis, and resource allocation.

Conclusion
There is no question that AI is gaining popularity and will become part of every company’s continuous improvement to maintain their competitive position. The time saved by allowing AI to assist in repetitive or routine tasks can be redeployed to other activities. However, implementation of AI is not free of risks and challenges. Knowing when and how to deploy AI practices are critical to avoid abuse of AI systems and to gain employee acceptance.

Bibliography
Brynjolfsson, E., & McAfee, A. (2017). The business of artificial intelligence: how AI fits into your data science team. Harvard Business Review, 98(4), 3-11.
Dalcher, D. (2022). The quest for artificial intelligence in projects, Advances in Project Management Series, PM World Journal, Volume XI, Issue III, March.
Grace, K., Salvatier, J., Dafoe, A., Zhang, B., & Evans, O. (2018). When will AI exceed human performance? Evidence from AI experts. Journal of Artificial Intelligence Research, 62, 729-754.
Gupta, C. (2020). Artificial Intelligence (AI) Influence in Project Management; PM World Journal, Vol. IX, Issue II, February.
Marr, B. (2020). Tech trends in practice: The 25 technologies that are driving the 4^th industrial revolution. Chichester: John Wiley & Sons.

Ong, Stephen; Uddin, Shahadat. Data Science and Artificial Intelligence in Project Management: The Past, Present and Future. Journal of Modern Project Management. Jan-Apr2020, Vol. 7 Issue 4, p1-8. 8p. DOI: 10.19255/JMPM02202.

Wang, Q. (2019).  How to apply AI technology in Project Management, PM World Journal, Vol. VIII, Issue III (April).

Dr. Harold Kerzner, Ph.D.
Senior Executive Director, International Institute for Learning

Harold Kerzner, Ph.D. is IIL’s Senior Executive Director for Project Management. He is a globally recognized expert on project management and strategic planning, and the author of many best-selling textbooks, most recently Project Management Next Generation: The Pillars for Organizational Excellence.

Disclaimer: The ideas, views, and opinions expressed in this article are those of the author and do not necessarily reflect the views of International Institute for Learning or any entities they represent.