algorithm · Artificial Intelligence · Project Management

What can project managers do about AI risks?

Since the launch of ChatGPT the interest, enthusiasm, and concerns about artificial intelligence  (AI) have grown dramatically.  According to ChatGPT Statistics and User Numbers 2023 – OpenAI Chatbot (, ChatGPT had a million users in its first week and 100 million as of August 2023. McKinsey estimates that 40 percent of business will increase their investments in AI (McKinsey Global Survey on AI, 1,684 participants at all levels of the organization, April 11–21, 2023; The state of AI in 2023: Generative AI’s breakout year | McKinsey). 

On November 17, 2023 we had a project management (PM) Stammtisch session on “Managing Artificial Intelligence (AI) – The Ethical and Social Implications.”  All participants indicated that AI solutions, specifically generative AI solutions, were being implemented in their organizations. One of the key questions from the audience was “what can we do as project managers about the ethical implications of AI in our organizations?” We have thoughts on that topic. 

First, it is important to understand the project life cycle for how AI systems are implemented. That understanding gives some indication on how AI development projects can impact the society and in which ways.  We wrote about AI life cycles here. Stakeholder roles in artificial intelligence projects – ScienceDirect 

Next, we wrote about the AI success factors here Artificial Intelligence Project Success Factors—Beyond the Ethical Principles | SpringerLink and AI accountability here jem.2022.44.18 ( Further, in our 2022 year-end blog, we summarized the aspects project sponsors, project managers, & team members implementing AI systems should consider to be accountable for the harms (minor to serious, intentional or unintentional) or benefits of the systems they develop. You can find that summary here: This was maxmetrics 2022 – maxmetrics.  

Now, for a couple of specific actions project managers can take to influence how AI solutions are developed in their organization. 

  • Scope. Establish the relationship between the project and society in the scope definition document. The scope definition document, or problem statement, is a contract that reveals the system’s goal and the behavior that can be anticipated. It defines the aims and rationale for the system and establishes the moral issues, and all aspects of the project impacted by the system’s context (e.g., country, industry sector, functional topic, and use case). It should explain how the system will impact society, including all stakeholders and the environment. 
  • Stakeholders. Engage representation for passive stakeholders in the project. The development team is the only group that can consider the concerns of all stakeholders. Thus, project managers should plan for, budget, and introduce inter-organizational solutions to include representatives of passive stakeholders. Passive stakeholder are affected by the project but may not have a direct benefit; they include end-users, local community, data subjects (sources from whom data will be collected), decision subjects (people that will be impacted by the AI), and workers (whose job may be impacted). 
  • Practices. Define/emphasize ethical practices. Establish an ethical function that includes policies, training, and an ombudsman or a whistle-blower process for project team members to voice their concerns. 
  • Investigations. Establish methods to investigate impacts and assess system risks. Algorithm auditing is a method that reveals how algorithms work and determining issues that should not arise. Algorithmic impact assessments investigate aspects of the system to uncover the anticipated impacts of the systems and to propose steps to address any deficiencies or harm. Risk assessments identify the potential implications and risks of the system, including legality and compliance, discrimination and equality, impacts on basic rights, ethical issues, and sustainability concerns. Specifically, the risk assessment provide an opportunity to identify and decide how to manage the risks, e.g., avoid, accept, or change the likelihood of the risk occurring.