With AI, can we make work human again?

AT THE National Day Rally 2025, Prime Minister Lawrence Wong framed artificial intelligence (AI) as a core economic engine for Singapore. The 2025 Budget followed suit, earmarking up to S$150 million for a new Enterprise Compute Initiative to help companies access advanced AI tools.

Yet, a paradox is emerging. Enterprise spending on AI is declining in Singapore – to 11.5 per cent of technology budgets this year, down from 15.5 per cent in 2024, according to ServiceNow’s Enterprise AI Maturity Index 2025 report. For a country lauded for its digital ambition, this raises an uncomfortable question: Why isn’t AI delivering the outcomes Singapore expects?

One concerning possibility is that Singapore’s investment in AI might not be addressing the right problems. Millions spent on training programmes and digital transformation initiatives have left many organisations feeling unprepared for AI.

AI isn’t ready for workers?

That has made them more cautious about going all in with the very technologies Singapore is promoting: Only 26 per cent of organisations surveyed describe themselves as AI-mature.

What if the AI paradox isn’t because workers aren’t ready for AI but because AI isn’t ready for workers?

The problem is not capability but adaptation. AI solutions still demand that humans bend their workflows to fit technology, instead of the other way around. Even in a country where labour productivity continues to rise, misapplied AI risks eroding these hard-won gains and creating very real economic costs.

Consider the average network operations manager at a telecommunications company. Each morning, the person juggles multiple disconnected systems for alerts, tickets, maintenance schedules, billing and analytics.

When a customer reports dropped connections, the fault team logs it manually but the customer relationship management doesn’t sync with network diagnostics.

Engineers are dispatched, yet updates to customers are delayed – if they arrive at all. Every small fault still relies on human stitching, despite years of “digital transformation”.

This is where many AI deployments fail. AI can automate tasks, but it cannot fix broken workflows, replace judgment, or interpret organisational complexity. Many deployments deliver dashboards and reports – speed without context, data without insight – leaving projects stalled, pilots unscaled, and workers frustrated.

Adapting to the worker

Now imagine AI truly adapting to the worker. Rather than navigating five separate applications, the manager could interact with a single conversational interface linking them all.

A voice assistant could retrieve diagnostics, update ticket statuses, and guide troubleshooting in plain language. Decision-making would no longer be a chore of switching contexts; it would feel intuitive, contextual, and human – again.

The economic stakes of this disconnect are clear. In Singapore, customer service agents often navigate more than three disconnected systems to resolve a single query. Although agents estimate complex issues could be resolved in 30 minutes, customers wait nearly five days, according to ServiceNow’s 2025 Customer Experience Report.

This hidden productivity drain – estimated to total 40 million hours on hold and S$1.3 billion in lost wages nationwide – directly threatens Singapore’s competitiveness as regional peers accelerate their own AI maturity. Worse, misapplied AI does more than disrupt operations; as seen in a recent high-profile case of fabricated research for the Australian government, poor oversight can quickly erode public trust and undermine national ambition.

The solution is not more dashboards or more metrics – it is their death. The future of work will demand invisible interfaces and conversational AI, where technology meets workers where they are – think Jarvis from the Iron Man franchise, acting as an assistant and adviser in complex situations. What was once science fiction is now the next step in operational design.

For enterprises seeking to unlock these outcomes, three rules stand out:

• Start with people, not models. Map how people really work across teams and processes before adding AI layers, so the technology supports real human needs rather than forcing people to adapt.
• Design for empathy, not dashboards. Connect siloed information into a central nerve centre that anticipates needs and supports decisions. Ensure AI can interpret intent and interact with people conversationally: this frees workers to focus on judgment, creativity and problem-solving instead of manual coordination.
• Measure human outcomes, not just pilot metrics. Evaluate success by improvements in productivity, customer satisfaction, and decision-making.

Singapore has the infrastructure, skills and capital to lead in AI. What it must now master is the human dimension. By putting people at the centre of AI – designing systems that work invisibly, conversationally and contextually – organisations can reclaim lost productivity, realise the promise of digital investment, and strengthen national competitiveness.

AI will not fail because Singaporeans are unprepared. It will fail if businesses continue to treat humans as extensions of technology instead of partners in it. The future of work is not about dashboards, metrics, or complexity. It is about making work human again and making AI work for people, not the other way around.

The writer is APJ innovation officer, Singapore at ServiceNow