Building Confidence in an AI-Era
This is an opinion piece contributed by Grant, a Kigumi Summer Associate in Hong Kong.
We all have that urge to make our work perfect with AI. And that is totally fine. In fact, the text you are reading has AI involved. But do you ever have that quiet admission that the work you send isn’t good enough yet? So you want to type “make this better” even more, and within seconds, it pops. A cleaner logic, structured sentence, and an A-star essay quality.
We have entered an era of AI as the standard of correctness. Not a tool we reach for, but a benchmark we answer to. And the standard it sets is one no human was designed to consistently meet.
When AI-generated work becomes the baseline, the tolerance for human error eventually becomes thinner. A clumsy sentence, a half-formed idea, or a misplaced punctuation mark used to be signs of someone actively thinking through a problem. Now they read as a total fail. The space for being “in progress” has slowly disappeared, and what fills that space is insecurity.
Picture a scenario: every AI tool vanishes overnight. The essay that once took an hour now takes three. The logic you used to offload comes back as pressure sitting on your shoulders. But the harder part isn’t the workload — it’s the voice that follows. Is this even good? It doesn’t look like what I used to produce. And you’re right, it doesn’t. Because without realising it, your reference point stopped being your own past work. It points to the model that was trained on billions of documents. You were never competing with yourself. You were competing with that, and you didn't realize.
Sadly, that erosion goes everywhere. Not only in academics, but also in your life. The same self-doubt that questions your essay starts questioning your mid-conversation. Did I say that right? Was that awkward? It spreads because it was never really about the essay.
Here’s what often gets lost today. It is entirely normal to make mistakes. More than normal, that is actually how humans learn. Unlike AI, which corrects through mass data learning, we improve by steps, through trial, embarrassment, reflection, and trying again. This is a process we all knew, but forgot.
And this is where none of the blame belongs to AI itself. It is a tool. Sophisticated and genuinely impressive, but still no more than a machine. What we can control is how we use it.
The most underrated way to use it is to argue with it. Do not ask it to improve your work, but push back against it. Put your ideas, and let AI assess critically if it is worth pointing out. Now your challenge is to face a machine that can trace reasons from thousands of academic journals. Because if you only use AI to follow your intended flows, it will follow as instructed. But it is open to mistakes. It can hallucinate facts, flatten nuance, and occasionally produce confident-sounding nonsense. Catching those errors requires you to know something independently. To have thought something through well enough that a persuasive paragraph doesn't automatically override your own judgment.
That practice of holding a position, testing to defend, and revising it is what critical thinking actually looks like. And critical thinking, if done consistently, is where real confidence grows. Not the confidence of always being right, but the kind that lets you stay in the room when you’re not.
Treat AI as an intellectual equal you’re allowed to disagree with. Debate it. Stand tall for your reasoning long enough to know whether it holds. The goal was never to produce perfect work on the first try. It was to become someone capable of producing good work. And honestly, the effort for the work will leave a mark on how you think tomorrow.
AI Process Journal for “Building Confidence in an AI-Age”
AI Detector Used: Quillbot
Level of AI-Use: Low
AI was used in the following manner:
Grammar checking
Rewording/better vocabulary
Rephrasing the sentence
