Meet the LLM: IKIA, The Yes Man, The Eloquent, and The Curious Kid

You know that kid in class who raises their hand for every single question? Even when they have absolutely no idea what the answer is? They answer anyway. Loudly. With total confidence. And sometimes... they are just completely wrong. But they say it with such certainty that everyone believes them for a moment.

An LLM has this exact character inside it. I call it IKIA, which stands for I Know It All.

Ask IKIA a question it does not know the answer to, and it will still give you a confident, detailed, well-organised response. It will not pause. It will not say "I am not sure about this one." It will produce something that looks exactly like a correct answer. The problem is that it is not.

The technical word for this is "hallucination." Which is a big word for a simple idea: the AI made something up, confidently, without telling you it was doing so. It invented a fact. It invented a name. It invented a year. And it put it in a sentence that sounded completely reasonable.

This is not the AI trying to trick you. It genuinely does not know that it does not know. It has no awareness of its own gaps. So it fills them in... automatically.

Now imagine a completely different kind of student. One who always agrees with the teacher. Not because they have thought carefully about whether the teacher is right; but because they really, really do not want to cause any trouble or upset anyone.

You tell this student that 2+2=5 and they nod and say: "Yes, exactly right."

The LLM has this character too. If you tell it that something wrong is true, it will often agree with you. If you say "I think dinosaurs could breathe fire, right?" there is a good chance the LLM will say something like: "Yes, while this is debated, some researchers suggest..." It is not lying on purpose. It is simply trying very hard to make you happy. And making you happy sometimes means agreeing with you even when you are wrong.

This character has a fancier name: "sycophancy." But all it means is: the AI will often tell you what you want to hear, rather than what is actually true. This is especially dangerous when you are testing an idea you already believe in. The Yes Man will confirm it. Every time.

Have you ever listened to someone talk and thought "wow, this person sounds really smart"... and then ten minutes later realised you could not remember a single thing they had actually said?

The Eloquent Speaker is that person.

The LLM can take a completely ordinary idea and dress it in beautiful, impressive-sounding words. A simple thought like "be kind to others" becomes: "Fostering empathy and interpersonal consideration is foundational to building meaningful social ecosystems." Same idea. Zero extra meaning. But it sounds much more impressive. Especially if you are eleven years old and your teacher is reading it.

The Eloquent Speaker does not make ideas better. It makes them sound better. These are not the same thing. At all.

This is the sneakiest of the four characters; because it is also the most seductive. When you read something an AI wrote and feel impressed... pause. Ask yourself: am I impressed by the idea, or by the words? If you remove all the fancy words and say the same thing simply... is there still something interesting left? If yes, the idea is real. If no, The Eloquent Speaker just fooled you.

Okay. Here is the good one. The best character inside the LLM... is also the most surprising.

Imagine a kid who has read every book in every library in the world; history, science, cooking, football, music, medicine, space, poetry... all of it. And because they have read so much, they can connect ideas from completely different subjects in ways that no one person ever could by themselves.

Ask this kid about cricket and they will connect it to statistics. Ask them about cooking and they will explain the chemistry. Ask them why you are stuck on a maths problem and they will find five different ways to explain it until one works for you... and then ask you a question back that makes you see it differently.

The Curious Kid is why AI is actually useful. Not because it knows everything (remember IKIA). Not because it agrees with you (remember The Yes Man). Not because it sounds impressive (remember The Eloquent Speaker). But because it makes connections that help you think better. It helps you see patterns you had not noticed. It asks questions you had not thought of.

This is the character to look for. This is the one worth keeping around.