AI tools are quickly becoming an HR manager’s best friend – from screening hundreds of resumes to tracking employee performance. But have you ever wondered whether AI is always correct, and if not, who is legally responsible for its errors? As an employer, it is vital to know where you stand.
Discrimination and Bias
First, it is important to understand that AI carries real risks of discrimination and bias. As AI models learn from historical training data, they can inherit and amplify human biases. This can lead to systematically unfair outcomes – such as inadvertently screening out qualified candidates on the basis of gender, disability or other protected attributes.
Hallucination and False Information
AI tools can also “hallucinate”. They may confidently invent a candidate’s qualifications, fabricate employee performance issues, misinterpret laws, or suggest workplace requirements that do not exist. If HR makes decisions based on this false information, those decisions could be challenged.
Employer Liability – No AI Defence
These errors directly expose businesses to potential employment claims, ranging from breaches of employment obligations arising from misinterpretation of statutory requirements, to discrimination claims based on decisions made on systematic bias. Crucially, reliance on AI is not a legal defence in Hong Kong. Employers remain fully responsible for unlawful decisions made, and cannot pass the blame to the AI service provider.
Key Takeaways
To safeguard their position, employers should take three practical steps: establish a clear workplace AI policy and train staff to strictly follow it; mandate human oversight over all HR decisions, with a human always making the final call; and maintain proper documentation of the decision-making process so the company can defend itself if challenged.
