Skip to content
RecTalk home

RecFest UK 2026: Hung Lee on AI, the Future of Talent Acquisition & Keeping Work Human

30 minWatch on YouTube (opens in a new tab)

About this episode

Nitin Sharma and Sean Kennedy speak with Hung Lee, a prominent voice in recruitment and talent acquisition, at RecFest UK. The conversation explores AI's impact on talent acquisition and challenges several industry assumptions about how technology should reshape work and organisations. Hung discusses why skills-based hiring may be anti-human, questions the need for extensive documentation, and examines the risks of using AI solely to extract more productivity from employees.

A central theme is Hung's "keep what you kill" approach to AI adoption: when technology makes someone more productive, should employees benefit from the time savings rather than the company capturing all gains? This could improve AI adoption, autonomy and reduce burnout. Hung also reimagines the future of TA, arguing the function won't disappear but will evolve as businesses work with AI agents and alternative capacity sources. Throughout the episode, he emphasises protecting human value—relationships, judgement, creativity and community intelligence that machines cannot replicate.

Key takeaways

  • Organisations should consider sharing productivity gains from AI with employees to encourage adoption and reduce burnout rather than extracting all benefits for the company.
  • Talent acquisition will evolve significantly as AI agents and non-human capacity reshape the scope and professional identity of TA roles.
  • Human skills like judgement, creativity, relationships and community intelligence remain irreplaceable and require active protection in an AI-driven recruitment landscape.
  • Skills-based hiring approaches may overlook human factors, whilst excessive documentation can hinder effective organisational functioning.

Keep watching

Be a guest

Think you should be on RecTalk?

If you've built, led or changed something in recruitment, we want to hear the unfiltered version.