A new divide at work: The AI fluent and the AI fearful – Opinion

Imagine this: A young analyst finishes her client report in 20 minutes using AI.

Her manager, twice her age, spends two hours on the same task. Both produce good work, but only one leaves early.

This small difference foreshadows a larger divide in Singapore’s working future. Some will be “AI bilinguals” – fluent in the “second language” of AI. Others will still be learning the alphabet.

The language of work

When Minister for Digital Development and Information Josephine Teo spoke of creating “AI bilinguals”, she described those who can think in two tongues – the mother tongue of professional expertise, and a new national language of AI.

Just as English once connected Singaporeans to the world and generated business opportunities, this new fluency promises to connect human expertise with AI, transforming work and improving economic outcomes.

The problem is, not everyone learns at the same pace. This new form of bilingualism could split society into those proficient in AI and those left behind.

Younger professionals seem to pick up this language effortlessly, through trial and error, much like learning a second language by immersion. A Google-Harris poll in late 2024 found that 93 per cent of Gen Z and 79 per cent of millennial office-based workers use two or more AI tools per week.

Older workers hesitate; not from inability, but from fear of being judged when they fumble. Many worry that using AI might make them look unskilled, slow, or even replaceable. Yet, they hold deep institutional knowledge and the professional “mother tongue” that keeps organisations running.

Companies like Microsoft and HP Singapore have recognised and addressed this learning gap with “reverse mentoring”, which involves pairing younger employees with senior leaders in an exchange of digital and leadership skills. IBM, a pioneer of this strategy since the early 2000s, first used it to bridge generations across social media – and now, across AI.

These efforts remind us that curiosity beats seniority. And that experience without renewal risks irrelevance.

A similar divide runs through the economy. Big firms experiment freely; small ones cannot afford to. DBS now runs more than a thousand AI and machine-learning models; OCBC processes six million AI-driven decisions daily, up from four million one year prior.

Meanwhile, only 14.5 per cent of small and medium-sized enterprises (SMEs) now use machine learning, and 77 per cent remain stuck in pilot mode. Surveys by Infocomm Media Development Authority (IMDA) and Enterprise Singapore show the hesitation comes from a lack of in-house skills. There also lurks the fear of wasted investment and of getting it wrong. But then these firms may get left behind, without the expertise to catch up.

Yet the national momentum is unmistakable: IMDA estimates that AI-driven growth pushed Singapore’s digital economy to 18.6 per cent of GDP in 2024, with firms reporting broad productivity and cost gains. The tide is rising — but not all boats are lifting at the same speed.

The paradox of AI fluency

What began as a digital divide is fast becoming a cognitive inequality: between those who think with AI (collaboratively), those who think through it (dependently), and those who think without it (autonomously), shut off from its use by choice or circumstance.

This is much like how Singapore’s bilingual policy once separated those fluent in English from those still rooted in dialects.

When English became the working language in the 1970s, it opened global doors but closed others at home. Dialect-speaking students struggled; later generations dismissed dialects as backward.

Just as the dominance of English once eroded our mother tongues, the rise of AI now risks eroding something equally vital – independent thought and nuanced understanding.

Some of us are using AI to become more efficient – at the expense of critical thinking.

Psycholinguist Norman Segalowitz’s distinction between functional and cognitive fluency captures our present dilemma.

Applied to AI, functional fluency is the ability to perform tasks, like crafting effective prompts. Cognitive fluency, however, means understanding how an AI-generated answer is made, what it omits, and when to question it.

One opens doors to efficiency, the other keeps them open by sustaining critical thinking.

Research bears this out. Studies from MIT and Accenture found that when workers were prompted to review AI outputs critically – adding “speed bumps” into their workflow – their accuracy rose without significantly increasing completion time. The best users take time to pause, reflect, recalibrate. Herein lies the insidious paradox. Mastery demands that we think with AI: fluent in its logic, alert to its flaws, ready to override when needed. Yet, the more fluently we think with AI, the harder it becomes to think against it.

Building inclusion into fluency

Singapore’s first bilingual policy took half a century to balance results and inclusion. We don’t have that luxury this time. AI evolves by the day, not decades.

To prevent a new divide, workplaces must normalise learning in public. Slack’s Fall 2024 Workforce Index found 45 per cent of Singapore workers are not comfortable admitting AI use to managers. Yet, the quiet adopters – the ones tinkering after hours – often learn the fastest.

If large firms are already fluent in AI, SMEs need “interpreters”, not just incentives. Programmes such as IMDA’s Chief Technology Officer-as-a-Service show what “translators” can look like – consultants who adapt AI to each sector’s “accent”. Shared data platforms like SGTraDex lower costs by pooling infrastructure. Fluency grows fastest when translation meets trust.

And schools must go beyond teaching how to use AI, towards teaching how to reason with it. The next generation will need not just digital literacy, but also moral clarity – the ability to question, to detect bias, to decide when not to trust.

Singapore’s first bilingual project was about survival: learning to speak two tongues to stay connected to the world. The next will be about conscience – learning to speak with machines without surrendering our own voice, or the sovereignty of our thought. Fluency was never just about words. It was always about understanding.

Daniel Chan is assistant dean (Office of Programmes) at the Faculty of Arts and Social Sciences, and senior lecturer (French) at the Centre for Language Studies, National University of Singapore.

Source: The Straits Times © SPH Media Limited. Permission required for reproduction.

ADV: Research Position at Centre for Banking and Finance Law Pair linked to private investigation agency charged over unlawful surveillance of cop