Is AI making us artificially intelligent, but unable to analyse?

Let me paint a picture.

You find yourself staring at a blank page. Not unusual. It’s part of the job. The difference this time is that you have a very capable assistant sitting quietly in another tab, waiting to help.

So you do what seems normal today. You type in a prompt. Something along the lines of, “Write me a sharp, insightful piece on artificial intelligence and its impact on human thinking”. In seconds, you have something that is not bad. In fact, it is quite good. Structured, clear and confident. Almost as if it were written by someone who knew what he or she were doing.

The only problem, of course, is that you didn’t write it.

And you know that to people who know you, who have read your work, heard you speak, or spent enough time with you, it will be apparent that you didn’t write it.

Then comes the moment of choice. Do you publish it anyway?

I suspect more and more people are saying yes, which raises a far more uncomfortable question than whether AI is intelligent or useful – that debate is already tired. Is artificial intelligence making us artificially intelligent? Or, at least, artificially making us appear more intelligent than we are?

This is no longer a fringe concern.

Microsoft’s 2024 Work Trend Index found that 75 per cent of knowledge workers globally now use generative AI at work. Gallup data suggests nearly half of all employees use it at least occasionally. In Singapore, the Infocomm Media Development Authority reports that three in four workers are already using AI tools regularly. The tools are here. The adoption is real and is a good thing.

The question we are not asking loudly enough is: “What are we doing with them?”

Let’s consider an analogy from the world of aviation. As autopilot systems became sophisticated enough to handle most phases of flight, pilots were still trained rigorously, still highly capable, but they were flying less and managing systems more. Over time, flying was no longer second nature to them and when things went wrong, some pilots, particularly those who had spent their entire careers in the automation era, struggled to take over manually. Not because they weren’t smart enough, but because they hadn’t been doing it enough.

Skill, it turns out, is not something you have. It is something you keep doing.

This generative AI moment feels like that. It doesn’t just give us answers. It gives us answers that sound like they’ve come from someone who has already done the thinking. And if we are not careful, we start to mistake access to thinking for thinking itself. We generate opinions without grappling with the issue. And slowly, almost imperceptibly, we start to look, and feel, more intelligent than we actually are. Which, if we’re honest, is quite appealing.

When students outsource answers

I have tried it too. There is a growing folder on my computer filled with beautifully written drafts I had very little to do with. Occasionally, I read them back and think: That’s quite sharp. Or worse, that actually sounds like me. Followed almost immediately by: I have no idea if I actually believe this.

Because intelligence is not just about producing good answers. It is about forming a point of view, and there is no faking that.

Nowhere is this tension more consequential than in education.

Across the world, educators are beginning to see the same pattern. Students are submitting work that is structurally sound, well-argued and polished, but when asked to explain it, defend it or apply it, the depth isn’t there. The output looks intelligent. The thinking behind it often isn’t more than skin deep.

Singapore’s students, operating in one of the most competitive academic environments in the world, feel this pressure acutely. The temptation to use AI shortcuts is high, and so are the stakes.

The irony is hard to miss. The very students working hardest to demonstrate capability may be outsourcing the capability they are meant to develop.

Humans think, AI assists

Let me pause here to say emphatically that this is not an argument to restrict AI. Quite the opposite. It is an argument to focus much more sharply on the skills that AI cannot replace, and the ones we must now actively develop. AI literacy should be structured into education from as young an age as possible. Not just how to use it, but how to use it well.

The real shift is not technological. It is cognitive. The question is no longer: “Can you produce a good answer?” Instead, it is: “Do you know what a good answer looks like and why?”

That requires a different set of capabilities. In business and management education, it is the ability to frame problems properly, to ask the right question before getting to answering them. AI can generate 10 answers in seconds. It cannot tell you if you are solving the wrong problem.

In engineering and the sciences, it is judgment under constraints, knowing which variables matter, what can be approximated and where precision is non-negotiable. AI can compute. It cannot decide what is worth computing.

In the humanities and social sciences, it is discernment, the ability to weigh sources, detect bias, and understand context. AI can synthesise information. It cannot truly understand meaning.

In healthcare and law, it is about making decisions where consequences matter and trade-offs are real. AI can assist with analysis. It cannot own the outcome.

Across all disciplines, one capability rises above the rest, the ability to think through something, and not just about it. This is what the man-machine pairing should look like. AI handles scale, speed and synthesis, humans handle judgment, context and consequence.

Used this way, AI becomes a force multiplier. It helps you see more, test faster and explore wider. Used poorly, it becomes a substitute for the thinking and the risk is that we become more convincing without becoming more capable.

I was recently at a talk where a senior Cabinet minister spoke about building his own AI agent using a Raspberry Pi and Claude to provide him with accurate, concise and relevant updates on demand. It was not presented as a novelty. It was presented as a necessity. AI, clearly, is now front and centre.

This is reinforced by the Government’s roll-out of Pair, a secure GPT platform for civil servants aimed at boosting productivity across the public sector. It will be interesting to see how it is used. As a sounding board? As an assistant? Or, quietly, as a ghostwriter?

Some universities globally are beginning to shift in this direction, placing greater emphasis on oral examinations, applied problem-solving, and iterative work that reveals how students think, not just what they produce. Not because they want less technology, but because they recognise that human capability must evolve alongside it.

There is a broader implication here, particularly for Singapore. For decades, cognitive capability, the ability to analyse, synthesise and communicate, has been the currency of value, especially among professionals, managers and executives. AI is now flattening the appearance of that capability. Well-structured sentences, logical flow and confident tone used to signal that someone had done the work. Now they are the default output of a machine.

If everyone can produce work that looks intelligent, what distinguishes those who actually are? The answer, I suspect, is that the signal has moved. It is no longer in how well something is written. It lies in how well something is thought through.

And that is much harder to fake.

So the next time you are staring at that blank page, sit with it a little longer. Use AI to teach you about the topic or to take care of something else mundane so you can dedicate a little more time to that blank page.

The ideas may come slower, they may be rougher, but they will be yours and for now, that still feels like the point of it all.

 

Krishnan Menon is a senior marketing and transformation leader with more than 30 years of experience across Asia-Pacific. He is also a distinguished fellow at the Nanyang Centre for Marketing and Technology at NTU’s Nanyang Business School.

Source: The Straits Times