KUALA LUMPUR (Aug 20): Malaysian firms are embracing artificial intelligence (AI) but the spending on the technology has yet to translate into meaningful productivity gains, according to a survey.
A study by the Malaysia Productivity Corporation of 57 manufacturing firms found that AI-related investment had only a weak relationship with labour productivity, suggesting that adoption by itself is not enough to make businesses more efficient.
The real payoff, according to the government agency’s Productivity Report 2026, depends on whether companies also redesign workflows, strengthen their data systems, equip workers with the right skills and improve management practices.
The report is published annually to track labour productivity and efficiency as well as to make recommendations. The agency, supervised by the Ministry of Investment, Trade and Industry, is mandated to develop and drive national productivity across industries and sectors.
Companies globally have poured billions into AI but the rapid investment and adoption have outpaced measurable gains in output, creating what has been called the “AI productivity paradox” by researchers.
“This is not evidence that AI is ineffective,” the agency said in the report. Rather, firms may end up adding cost and complexity if they introduce AI tools while retaining old processes, fragmented data and existing layers of approval, the report noted.
Many Malaysian firms are still at an early stage of turning AI adoption into measurable performance improvements, with less than 30% reporting a high level of readiness to integrate the technology effectively.
Still, companies are beginning to see benefits. Nearly three-quarters of firms surveyed reported some improvement from AI, particularly in areas such as compliance, customer engagement and innovation.
The agency outlined measures to address the AI productivity paradox, including better measurement of AI's impact, redesigning jobs and accelerating reskilling and improving coordination and accountability across ministries and agencies.
It also recommended prioritising high-impact sectors and small-to-medium businesses, developing practical AI guidelines based on real business problems and measurable outcomes, and establishing an AI productivity research network linking government, universities and industry.