Understanding Stakeholders
Key stakeholders must be involved in value creation and economic profit creation throughout the design, development, and deployment of the AI system. It is, therefore, wise to begin AI design by identifying any group or individuals who can affect or are affected by the AI system. In this stakeholder identification guide, we will cover:
- What does stakeholder identification mean
- Why stakeholder identification is needed
- Process for stakeholder identification
- Tools for stakeholder identification
- Limitations and their consequences
What does stakeholder identification mean
A stakeholder can refer to one or more groups or individuals, such as customers, employees, suppliers, partners, distributors, debtholders, or anyone. Typical definitions of stakeholder from Bryson (2004) include:
‘All parties who will be affected by or will affect [the organisation’s] strategy’.Â
‘Any person, group, or organisation that can place a claim on the organisation’s attention, resources, or output, or is affected by that output’.
‘People or small groups with the power to respond to, negotiate with, and change the organisation’s strategic future’.
Identifying stakeholders refers to “who are you building the solution” and “who can affect or is affected by the AI system”. Download the ‘Who…’ Approach to Identifying Stakeholders.
Why stakeholder identification is needed
ISO 42001, a standard related to artificial intelligence management systems, emphasises identifying and engaging with all stakeholders impacted by AI systems. The standard aligns with AI ethics and governance principles, ensuring that AI systems are designed responsibly and deployed in a manner that respects human rights and fosters trust.
ISO 12791 recommends stakeholder identification in the system inception phase to ensure a comprehensive approach to mitigating bias throughout the AI lifecycle.
The EU AI Act emphasises identifying and engaging with all stakeholders affected by AI systems, including individuals, communities, and relevant interest groups. These best practices encourage continuous stakeholder input to refine the system and address emerging ethical or operational concerns.
Tools for stakeholder identification
To assist you in systematically identifying stakeholders, we offer free, easy-to-use tools. Simply click ‘Download Guidelines’ on the right to get started.
Free Resources for Stakeholder Identification
A ‘Who…’ Approach to Identifying Stakeholders
A ‘Why, how, where…’ Approach to Identifying Stakeholders tool
Mitigate and resolve 15 Types of Bias specific to your project with detailed guidance from problem definition to model monitoring. Packed with practical methods, research-based strategies, and critical questions to guide your team. Comprehensive checklists with +75 design cards for every phase in the AI/ ML pipeline Mitigate and resolve 15 Types of Bias specific to your project with detailed guidance from problem definition to model monitoring. Packed with practical methods, research-based strategies, and critical questions specific to your use case. Customised checklists and +75 design cards for every phase in the AI/ ML pipeline
Limitations and their consequences
Here are some significant limitations and their consequences (Salado & Nilchiani, 2013) for you to address in your design. Â
- The lack of common frameworks results in the use of classification or taxonomies to identify stakeholders’ roles and responsibilities.Â
- It may only be feasible to guarantee the inclusion of some relevant stakeholders.
- It may not be feasible to verify the accuracy or appropriateness of all stakeholders involved.
These three limitations lead to negative consequences in the development of AI systems:Â
- (from Limitation 1): Creativity in the stakeholder identification process is constrained by an overreliance on predefined categories, which leads to ” outside-the-box” thinking.
- (from Limitation 1): Stakeholder representation must be revised because rigid categorisation excludes broader behavioural diversity.
- (from Limitation 2): Requirements are incomplete as not all relevant stakeholders are identified or analysed.
- (from Limitation 3): Requirements are inaccurate because they are derived from stakeholders who may need to be more appropriate or relevant.
Sources
Bryson, J.M., 2004. What to do when stakeholders matter: stakeholder identification and analysis techniques. Public management review, 6(1), pp.21-53.
Salado, A. and Nilchiani, R., 2013. Contextual-and behavioral-centric stakeholder identification. Procedia Computer Science, 16, pp.908-917.