AI Governance for Projects & PMOs

Over the past 12 months, there has been a significant increase in the availability of AI tools for consumers. Large Language Models (LLMs) like Chat GPT and Microsoft Copilot have enabled more people to leverage these tools for enhanced productivity and efficiency. Many companies are eager to adopt AI technologies, and the McKinsey Global Survey on AI in 2023 indicated that ‘65% of respondents report their organisations are regularly using generative AI (gen AI) in at least one business function,’ representing a substantial rise from previous years. However, some organisations, including Apple, Samsung, and several major US banks, remain hesitant and have prohibited staff from using these tools, primarily due to security concerns.

Despite varying organisational stances, it is widely anticipated that most businesses will soon integrate LLMs in some capacity. In project management, early adopters are already utilising these tools for schedule management, risk management, data analysis, report development, and secretariat functions. The benefits can be significant for organisations willing to take the risk, as specialists and subject matter experts can now shift their focus to higher-value activities, improving productivity without sacrificing necessary repetitive or analytical work.

The Need for AI Governance

While the productivity benefits are clear, the use of AI also presents challenges. Most of the security concerns deterring some organisations today can likely be mitigated with secure, enterprise-level LLMs. Nevertheless, AI introduces several potential issues:

• LLMs can ‘hallucinate,’ providing inaccurate information, which in a project management context, could lead to data errors, incorrect narratives, and flawed decision-making.
• LLMs may be trained on biased data, potentially perpetuating and amplifying existing biases, leading to unfair treatment of certain groups.
• Although AI handles repetitive and data-intensive tasks, it is crucial to maintain a balance, encouraging team members to engage in critical thinking and problem-solving. There is a concern that ‘if you don’t use it, you lose it!’
• AI involves processing large amounts of data, raising concerns about data privacy and intellectual property rights.

Moving to AI as Business as Usual

For AI to become a widely accepted ‘business as usual’ tool, an appropriate governance framework will be essential for most organisations. This is particularly important in project management, where data often informs critical decisions, stakeholder communications, and ensures project compliance and accountability. Proper governance ensures ethical and effective AI use, maintains data integrity, and mitigates risks associated with AI-generated information through rigorous validation and oversight processes.

Key Principles of AI Governance for PMOs

Effective AI governance in PMOs hinges on four key principles:

1. Transparency: Clear documentation and communication of AI processes are essential, ensuring stakeholders understand how AI tools are used and how decision-making works with AI-developed data. This fosters trust and enables project managers to trace and explain AI-generated outputs.
2. Accountability: Establish clear roles and responsibilities for managing AI systems, ensuring there is a defined person or team responsible for the AI’s performance and ethical use.
3. Fairness: Prevent biases in AI models by using diverse datasets and regularly auditing AI systems to ensure they treat all project stakeholders equitably.
4. Privacy: Protect sensitive project data processed by AI tools, complying with data protection regulations, securing data storage, and ensuring personal and proprietary information is not misused.

Establishing an AI Governance Framework for the PMO

To effectively integrate and manage AI technologies within a PMO, it is essential to ensure that the correct level of management and technical oversight is in place. Establishing a PMO-level committee or task force will help align AI initiatives with organisational and PMO objectives, adhering to ethical and regulatory standards.

Components of an AI Governance Framework:

1. Establish an AI Governance Committee: This committee should include key stakeholders such as project managers, data analysts, risk managers, IT security professionals, and representatives from PMO leadership. This diverse composition ensures all aspects of AI implementation are considered.
2. Define Objectives: Clearly define the primary objectives of the AI governance committee, including the development of an AI strategy or Charter, developing AI usage policies, ensuring regulatory compliance, managing AI-related risks, and promoting ethical AI practices.
3. Regular Meetings and Reporting: The committee should meet regularly to review project status, assess compliance with AI policies, and address emerging risks or ethical concerns. Transparency about AI usage in decision-making forums is crucial for maintaining trust and accountability.
4. Establish Subcommittees: For larger projects and PMOs, subcommittees can focus on specific AI governance aspects such as data privacy, bias mitigation, or AI-driven project risk management, providing detailed oversight and recommendations.

Assigning Responsibilities for AI-Related Decision-Making and Oversight

Defining clear roles and responsibilities within the PMO is critical. A responsibility assignment matrix (RAM or RACI) can be useful. Key roles might include:

• AI Governance Lead: Oversees AI implementation and management, ensuring alignment with objectives, risk management, and ethical AI practices.
• Data Privacy Officer: Ensures compliance with data protection regulations, overseeing data privacy measures and conducting regular audits.
• AI Ethics Officer: Identifies and mitigates ethical risks, ensuring AI systems are fair, transparent, and free from biases.
• Project Managers: Use AI tools within project workflows, monitoring AI outputs, validating accuracy, and ensuring transparency.
• IT Security Team: Protects AI systems from cyber threats, implementing security protocols and conducting vulnerability assessments.
• Training/HR Team: Trains PMO staff on AI tools and ethical implications, maintaining high standards of AI literacy and ethical awareness.

AI governance will have tentacles into every part of an organisation so it will become critical to understand who has the accountability for which data sets and for maintaining data integrity.

Figure 1.0 AI Accountability Venn Diagram

Ensuring Compliance with Local and International Regulations

Compliance with local and international regulations is crucial. PMOs must ensure their AI implementations adhere to data protection laws such as the Privacy Act (1988) in Australia, GDPR in Europe, or the local equivalent legislation. This involves robust data handling and storage practices and ensuring that the team stay updated on evolving AI regulations and standards. Clear policies and training programs for regulatory requirements are essential to maintain compliance and avoid legal repercussions.

Building a Culture of Responsible AI Use

Promoting ethical AI usage involves integrating ethical considerations into all aspects of AI deployment. This includes training on AI ethics, leaders modelling ethical behaviour, establishing clear policies, fostering open communication, encouraging diverse teams, conducting regular audits, and maintaining transparency through reports and stakeholder engagement. By doing so, PMOs can ensure responsible AI use, optimising project outcomes and maintaining stakeholder trust.

Conclusion

The integration of AI into project management can offer substantial benefits for users, enhancing productivity and allowing specialists to focus on higher-value tasks. However, the potential risks associated with AI, such as data inaccuracies, biases, and privacy concerns, necessitate robust governance frameworks and mean that a move to incorporating AI enable tools into business-as-usual operations should not be taken lightly.

Establishing dedicated AI governance committees, defining clear roles and responsibilities, and ensuring adherence to local and international regulations are key steps in mitigating risks and fostering a culture of ethical AI use. By embracing these practices, PMOs can harness the productivity boosting power of AI while ensuring accuracy, compliance, project integrity and ultimately driving better project outcomes.

If we can help you with building your PMO AI Governance Charter contact us today!

Copyright Louise Gardner, 2024. All rights reserved.