世界经济论坛-以人为中心的人力资源人工智能（英）-2021.12-59页_5mb

Summary of Human-Centred Artificial Intelligence for Human Resources: A Toolkit for Human Resources Professionals

Core Content

This toolkit provides HR professionals with guidance on the responsible and effective use of AI-based HR tools. It outlines the key considerations, implementation strategies, and ongoing monitoring practices necessary to ensure that AI is used in a way that aligns with ethical, transparent, and human-centric values.

The document emphasizes that AI is not a one-size-fits-all solution, and its use in HR requires careful evaluation and planning. It highlights the importance of understanding both the capabilities and limitations of AI systems, particularly in relation to human judgment, bias, and the ethical implications of automated decision-making.

Main Views

1. The Role of AI in HR

• AI is being applied to almost every aspect of HR, from recruitment to employee engagement and retention.
• AI tools vary in their functions, but they generally fall into categories such as automation, augmentation, and prediction.
• AI is not expected to replace HR professionals entirely but rather to assist in specific, repetitive, or complex tasks.

2. How AI Works

• Most AI-based HR tools use machine learning (ML), which identifies patterns in training data to make decisions or predictions.
• ML systems require relevant and sufficient training data to function effectively.
• The inputs and outcomes of AI systems are crucial to understanding how they operate and what they are designed to achieve.

3. Balancing Human and AI Decision-Making

• Human decision-making is based on life experiences and common sense, allowing for nuanced understanding and empathy, but can be biased and slow.
• AI decision-making is based on training data, enabling speed, scalability, and pattern recognition, but lacks the ability to understand context or check for the "sensibility" of decisions.
• AI can amplify human bias, especially if the training data reflects existing prejudices.

Key Considerations

1. Bias

• AI systems can inherit and amplify biases present in training data.
• HR professionals must assess whether the data used to train AI tools is representative and free from discriminatory patterns.
• It is important to recognize that AI is not inherently fair or objective.

2. Data Privacy and Security

• AI tools often require access to sensitive employee data, which must be handled carefully to ensure compliance with privacy regulations.
• Organizations should evaluate how data is collected, stored, and used by AI systems.
• Transparency in data usage is essential to maintain trust.

3. Transparency and Explainability

• AI decisions should be explainable to ensure accountability and fairness.
• HR professionals must be able to understand and justify the outcomes of AI systems.
• Lack of explainability can lead to distrust and legal issues.

Implementation and Buy-in

• A multistakeholder assessment team is essential to evaluate AI tools effectively.
• The team should include HR professionals, IT staff, legal experts, diversity and inclusion specialists, data scientists, and employee representatives.
• The chief human resources officer (CHRO) and other decision-makers must be involved in the adoption process.
• A clear pitch should be developed to justify the use of AI, including cost, risk, and benefits.

Ongoing Maintenance and Monitoring

• AI tools must be continuously monitored for performance and ethical compliance.
• Regular reviews are needed to ensure that the system remains aligned with organizational goals and values.
• The system should be updated as needed to reflect changes in the workforce or business environment.

Tool Assessment Checklist

• Evaluate the purpose of the AI tool and whether it aligns with HR goals.
• Assess the source and content of the training data to ensure relevance and fairness.
• Consider the risk level of the tool and whether it requires additional oversight.
• Ensure the tool is transparent and explainable to stakeholders.
• Check for potential bias in the system and its impact on decision-making.

Planning Checklist

• Establish AI principles and policies specific to HR.
• Review existing governance structures to ensure compliance with legal and ethical standards.
• Document the decision-making process and create a log of actions taken.
• Plan for ongoing maintenance and monitoring of AI systems.
• Ensure buy-in from all relevant stakeholders and prepare for integration challenges.

Conclusion

This toolkit is a comprehensive guide for HR professionals to navigate the complex landscape of AI adoption. It encourages a balanced, ethical, and human-centered approach to using AI in HR, ensuring that the technology is used responsibly and effectively. The accompanying checklists provide practical tools to help organizations make informed decisions and maintain oversight of AI systems throughout their lifecycle.