Using SFIA Levels of responsibility to analyse what tasks/responsibilities to assign to AI

There seems to be value in this approach but it needs further exploration. Get in touch if you want to take part.

SFIA's established levels of responsibility framework provides a structured approach for organisations to evaluate and assign tasks between human professionals and AI systems.

• By understanding how AI capabilities align with SFIA's seven levels - particularly in terms of autonomy, influence, complexity and business skills - organisations can make more informed decisions about AI implementation.
• This systematic approach helps balance the opportunities of AI automation with appropriate human oversight, ensuring AI is deployed effectively while maintaining professional standards and managing risks.

1. Summary of SFIA Levels of Responsibility

SFIA defines seven levels of responsibility, each with varying degrees of autonomy, influence, complexity, and business Skills/behavioural factors required:

Level 1 - Follow

• Autonomy: Works under close direction with minimal discretion.
• Influence: Minimal influence; works within immediate colleagues.
• Complexity: Performs routine activities in a structured environment.
• Business Skills/Behavioural factors: Basic communication and digital skills.

Level 2 - Assist

• Autonomy: Works under routine direction with limited discretion.
• Influence: Interacts with immediate colleagues and may have external contact.
• Complexity: Performs a range of work activities in varied environments.
• Business Skills/Behavioural factors: Sufficient communication and digital skills for role.

Level 3 - Apply

• Autonomy: Works under general direction; uses discretion in identifying and responding to complex issues.
• Influence: Interacts with and influences colleagues; may oversee others.
• Complexity: Performs a range of complex and non-routine work.
• Business Skills/Behavioural factors: Effective communication and application of methods and tools.

Level 4 - Enable

• Autonomy: Works under general direction within a clear framework of accountability.
• Influence: Influences customers, suppliers, and partners at the account level.
• Complexity: Broad range of complex technical or professional activities.
• Business Skills/Behavioural factors: Fluent communication, risk awareness, and analytical approach.

Level 5 - Ensure, advise

• Autonomy: Works under broad direction; self-initiated work.
• Influence: Significant influence over resources and collaborative work.
• Complexity: Extensive range of complex technical and professional activities.
• Business Skills/Behavioural factors: Leadership, risk assessment, and strategic application of knowledge.

Level 6 - Initiate, influence

• Autonomy: Authority and accountability for a significant area of work.
• Influence: Influences policy and strategy formation; initiates influential relationships.
• Complexity: Highly complex work activities covering technical, financial, and quality aspects.
• Business Skills/Behavioural factors: Organisational leadership, risk management, and strategic innovation.

Level 7 - Set strategy, inspire, mobilise

• Autonomy: Authority over all aspects of a significant area of work.
• Influence: Influences industry developments and long-term strategic relationships.
• Complexity: Applies highest level of leadership to strategy and governance.
• Business Skills/Behavioural factors: Strategic management, innovation, and industry leadership.

2. How employers decide what tasks/responsibilities to assign to AI without SFIA

Without a structured framework like SFIA, employers typically rely on:

• Experience and intuition: Decisions are often based on the employer's or team's past experiences with similar technologies.
• Ad hoc analysis: Tasks may be assigned based on immediate needs or perceived benefits without a systematic approach.
• Trial and error: Employers may experiment with AI for various tasks to see what works, leading to inconsistent application.
• External advice: Consultants and industry experts may guide decisions, which can vary greatly depending on who is consulted.

3. How employers decide what tasks/responsibilities to assign to AI using SFIA

Using SFIA as a guide, employers can make more informed and structured decisions:

• Role matching: Employers can match AI tasks with the appropriate SFIA levels of responsibility, ensuring tasks are suitable for the level of expertise required.
• Skill assessment: By assessing the skills required at each SFIA level, employers can determine if AI can perform certain tasks or if human oversight is necessary.
• Task complexity: SFIA helps identify the complexity of tasks and match them with AI capabilities, ensuring AI is used where it can be most effective.
• Risk management: Employers can use SFIA to evaluate risks and ensure AI applications comply with organisational standards and ethics.

4. Guidelines for employers using SFIA to assign AI tasks/responsibilities

• Identify task complexity: Use SFIA as a reference point to assess the complexity of tasks. Assign routine, structured tasks to AI (SFIA Level 1-2) and complex, unstructured tasks to higher levels (SFIA Level 4-7).
• Match skills to AI capabilities: Evaluate the AI's capabilities against SFIA skill requirements. Assign tasks that match the AI's technical proficiency and ability to handle data securely.
• Evaluate autonomy requirements: Determine the level of autonomy required for each task. AI should handle tasks requiring minimal discretion and close direction, while tasks needing significant discretion should remain under human oversight.
• Consider ethical and security implications: Ensure AI tasks align with SFIA's focus on security, privacy, and ethics. Assign AI to tasks where these factors are well-managed.
• Use incremental implementation: Start with lower-level tasks (SFIA Level 1-2) and gradually move to more complex tasks (SFIA Level 3-4) as the AI system proves reliable.
• Regularly review and adjust: Continuously monitor AI performance and reassign tasks as needed, based on evolving SFIA levels and organisational needs.