Anthropic, the company behind Claude AI, published a study last week that we keep going back to. The paper, Labor market impacts of AI: A new measure and early evidence by researchers Maxim Massenkoff and Peter McCrory, looks at which jobs AI can theoretically do versus which jobs it actually handles today. It's the most concrete data we've seen on AI jobs impact in Malaysia and the wider region.
The short version: AI is already affecting white-collar jobs in Malaysia, particularly in shared services, BPO, fintech, and IT. The study shows AI can theoretically handle 90%+ of computer-based office tasks, but actual workplace adoption sits at roughly 33-40%. The biggest immediate signal is a 14% drop in hiring rates for entry-level workers in AI-exposed occupations.
AI can do way more than people actually use it for
For computer and math workers, AI can theoretically handle 94% of their tasks. In actual workplace usage? Only 33%.
That gap shows up across every job category. Office and admin roles: 90% theoretical, roughly 40% observed. Business and financial roles: similar story.
The researchers didn't rely on surveys or expert guesses. They used real conversation data from Claude, tracked what people actually ask AI to do at work, weighted fully automated tasks more heavily than assisted ones, and mapped it all against roughly 800 occupations.
They call this "observed exposure." It's a more honest measure than the doom-and-gloom headlines suggest.
Which jobs are most exposed to AI right now?
The study ranks occupations by how much of their work AI already covers:
| Occupation | AI task coverage |
|---|---|
| Computer programmers | 75% |
| Customer service reps | 70% |
| Data entry keyers | 67% |
| Market research analysts | 65% |
Meanwhile, 30% of workers have zero AI coverage. Cooks, bartenders, motorcycle mechanics, lifeguards. If your job needs hands more than a keyboard, AI isn't coming for it anytime soon.
Which jobs will AI replace in Malaysia?
Malaysia's workforce looks different from the US workforce this study measured. But the patterns carry over.
The most AI-exposed workers in the study earn 47% more on average, are more likely to hold graduate degrees, and are 16 percentage points more likely to be female. AI hits white-collar, office-based roles hardest.
For Malaysia, that points to shared services centres, BPO operations, fintech back offices, and IT departments. These are the sectors that grew fast in KL, Penang, and Cyberjaya over the past decade. They're also the most exposed.
The study found no systematic increase in unemployment among highly exposed workers. Not yet. But the researchers spotted a 14% drop in job-finding rates for young workers (ages 22-25) in exposed occupations since ChatGPT launched. People aren't getting fired, but companies are hiring fewer of them.
An Ipsos survey reported by Malay Mail found two in three Malaysians believe AI will bring more jobs in 2026. We hope they're right. But optimism without a plan is just vibes.
What Malaysian businesses should do about this
The gap between what AI can do and what it actually does at work is your opening.
Look at where AI already works well in your company. Customer service, data entry and reporting, code review, report writing. You don't need to wait for the theoretical 94% to show up. The tools for that 33% exist right now.
Your people matter more than your tools, though. If your junior staff spend most of their day on tasks AI handles well, train them on the stuff AI can't do. Client relationships. Strategy. Figuring out what the AI got wrong. Hiring is slowing for entry-level roles in exposed fields, so the people you already have are worth more than you think.
One more thing: think beyond internal efficiency. If a competitor uses AI to serve customers faster or ship features cheaper, your current setup starts looking expensive. The companies pulling ahead are the ones that build AI into their actual products, not just their back office.
What does the AI jobs impact in Malaysia actually look like going forward?
AI's impact on jobs is real but uneven. The gap between capability and actual usage is still enormous, and that gap is where Malaysian businesses have room to move. You get to decide how AI fits into your operations, on your terms, while the technology is still catching up to its own potential.
If you're working through where AI fits in your business, or you need help building AI-powered tools for your team, we'd like to hear from you. We've worked with Malaysian companies on everything from automated data pipelines to custom platforms.
