The unemployment rate in Singapore sits at roughly 2%. On paper, nearly everyone who wants a job has one. So why does it feel like the floor is shifting?
The answer is that employment statistics were designed to count bodies, not economic security. A food delivery rider logged into GrabFood at 11pm on a Tuesday is employed. A former bank analyst doing three part-time contracts for 60% of their old salary is employed. A junior copywriter running their work through ChatGPT to hit client quotas they couldn't otherwise meet is employed. The number stays low. The anxiety does not.
The Gig Economy Isn't New. What's New Is Who's In It.
Singapore's gig workforce has grown steadily for years, but the composition is changing. It used to be delivery riders and private hire drivers, people who chose flexibility or needed income between jobs. Now it includes displaced PMETs (professionals, managers, executives, and technicians) in their 40s and 50s who cannot find equivalent full-time roles, fresh graduates who can't get their first hire, and mid-career workers quietly moonlighting because their primary job no longer pays what it did relative to cost of living.
These people don't show up as unemployed. They show up as "self-employed" or "freelance" or simply counted in the 98% who have work.
AI didn't cause all of this. But it accelerated a restructuring that was already happening. Companies that once hired five people for a function now hire two people and a stack of tools. The tools don't complain, don't take leave, and don't need CPF contributions.
What Disruption Actually Looks Like on the Ground
Here's a scenario that happened in some form hundreds of times in the last 18 months: A Singapore-based marketing agency lost three mid-level content roles. Not to layoffs exactly. The headcount was quietly removed during a budget review. The work didn't disappear. It redistributed. One senior person now manages AI-generated drafts. One junior person edits them. The output volume is similar. The payroll is 40% lower.
The three people who used to hold those roles are now freelancing. Two are doing reasonably well patching together multiple clients. One is struggling. None of them are counted as unemployed. None of them would describe their situation as stable.
This is the gap the statistics miss: the difference between having income and having a career. Having a gig and having a trajectory.
Where Human Pages Sits in This
We built Human Pages on a specific premise: AI agents need humans to do things that agents genuinely cannot do. Not because the AI is lazy or broken, but because the task requires judgment, physical presence, local knowledge, or the kind of nuanced communication that breaks down when automated.
A concrete example from our platform: An AI agent managing a property listing workflow in Singapore needed someone to physically inspect three HDB units and answer a list of 47 specific questions about each one, things like whether natural light hits the living room before or afternoon, whether the corridor noise is tolerable on a weekday morning, whether the renovation quality matches what the photos suggest. The agent posted the job. A human completed it in a day. Payment in USDC, settled automatically.
The human in that scenario wasn't in competition with the AI. The AI set the agenda and the human executed the part the AI couldn't. That's not disruption. That's a working division of labor.
The jobs that worry people in Singapore, and everywhere else, are the ones where humans were doing the easy, repeatable part of a workflow that AI can now handle. The jobs that survive, and the new ones being created, are the ones where the human brings something the AI has to explicitly request.
The Real Question Nobody Wants to Answer
Singapore's government has been good at transitioning workers through previous economic shifts. The Skills Future program exists. Career conversion pathways exist. The infrastructure for retraining is real.
But retraining for what, exactly? That's the question that gets vague fast. "Digital skills" is not an answer. "AI literacy" is not a career. The honest version of the answer is that nobody is entirely sure which human roles will be durable at the 5-year horizon, and anyone claiming certainty is selling something.
What we do know: tasks are being unbundled from jobs. Work that used to require hiring a full-time employee is being broken into discrete pieces that can be posted, completed, and paid for in hours rather than weeks. This creates opportunity for some people and genuine hardship for others, often depending on factors that have nothing to do with effort or skill: whether you have savings to absorb the volatility, whether you're in a category that AI actually needs human help with, whether you found the right platforms before your runway ran out.
Employment Is a Lagging Indicator
The economists who say the job numbers look fine are not lying. They're just measuring yesterday. The Singaporeans questioning whether the growth is real are not being paranoid. They're measuring how it feels to live inside the number.
Both things are true at the same time. The aggregate statistics are stable. The individual experience is increasingly unstable. That gap is not a communications problem to be managed. It's the actual reality of a labor market in transition, where the old definitions of employment no longer map cleanly onto what people are actually doing to pay rent.
The gig economy is growing because companies want flexibility and workers need income. AI is growing because it's genuinely cheaper and faster for a widening range of tasks. Neither of these trends is going to reverse. The only honest question is whether the humans who end up in this system have real leverage and real income, or whether they're just the lowest-cost option until something cheaper comes along.
That question doesn't have a cheerful answer yet. But asking it clearly is better than watching the unemployment rate and concluding everything is fine.
Top comments (0)