Where should AI assistance end and human hiring decisions begin?

We recently launched OnSkillDemand, and one question keeps coming up: how much responsibility should AI have in the hiring process?

Our approach is to use AI for structured assistance—such as analyzing vacancy requirements, supporting interviews, and organizing evidence—while keeping people responsible for every hiring decision.

I’d be interested to hear from founders, recruiters, and candidates:

• Which parts of hiring should AI help with?

• Which decisions must always remain human?

• What evidence or explanations would you need before trusting an AI-assisted assessment?

We’re also exploring how a skills Knowledge Graph could connect roles, required competencies, interview evidence, and development paths without turning hiring into an opaque score.