Skip to main content

Singapore HR survey finds human approval in 85% of AI use cases

Larger employers reported greater adoption, while concerns included the potential erosion of HR skills and uncertainty over responsibility for AI errors.

By Arif Rahman

Computer and recruitment documents on an office desk
Photo: CNA

People retain decision-making authority in 85 per cent of the human resources AI applications recorded by a Singapore National Employers Federation (SNEF) survey published on October 30. The findings draw on responses from 1,000 Singapore employers surveyed during June and July, covering almost 3,000 applications.

The other 15 per cent allow AI to decide without prior human approval. Within that group, 73 per cent concern administrative tasks that are routine or handled at high volumes, such as screening resumes or assessing skills. SNEF said almost every application operating without sign-off nevertheless had someone assigned to step in and revise or reverse decisions.

Among employers whose workforce exceeded 200, adoption stood at 81 per cent. The figure was 65 per cent for small and medium enterprises with no more than 200 employees. Recruitment, employee learning and workforce planning attracted greater use, whereas departure-related matters attracted the least.

SNEF president Kuah Boon Wee told the inaugural SNEF Employers’ Summit that monitoring should be proportionate to the consequences, with decisions affecting livelihoods requiring closer scrutiny than routine work.

Potential weakening of HR capabilities through excessive dependence on AI concerned 39 per cent of respondents. Other concerns were deciding when professional judgement should take precedence, at 36 per cent; responsibility for AI mistakes, at 35 per cent; and inadequate training, at 31 per cent.

Speaking at the summit, Acting Manpower Minister Jasmin Lau said the government was reviewing support schemes to strengthen assistance for graduates and retrenched mid-career workers. She said policies would adjust incrementally because the scale of potential AI disruption remained uncertain.

Keep reading: AI

All AI stories

Latest in Technology

All latest