Hallucinations (பொய் சொல்லுதல் / Confident Mistakes)

Let's ask the confident Auto Driver! (கான்ஃபிடன்ட்டா பொய் சொல்ற ஆட்டோக்காரரை பாப்போமா!)

Technical Meaning: பொய்த்தோற்றம் (Poythotram) - When an AI confidently generates completely false or fabricated information as if it were a fact.

The Core Idea

Because Large Language Models are just predicting the most statistically likely next word, they don't actually possess a human concept of "truth" or "facts." If an LLM doesn't know the answer to your question, its mathematical drive to complete the sentence often leads it to confidently invent fake facts, fake names, or fake code. This is called a Hallucination.

The Origin Story

Hallucinations are the biggest hurdle preventing AI from being used in high-risk fields like medicine and law. In a famous 2023 case, a lawyer used ChatGPT to write a legal brief, and the AI literally invented fake court cases and citations because they "sounded" like real legal language. Researchers are trying to solve this using techniques like RAG and Fact-Checking layers.

The Tamil Analogy

Imagine asking a confident Auto Driver in Chennai for directions to an obscure street.

The Auto driver actually has no idea where that street is. But because he wants to look knowledgeable, he will never admit he doesn't know.

Instead, he will confidently wave his hand and say, "Go straight, take a left at the signal, go past the temple, and it's on the right." He delivers it with absolute 100% confidence, even though he just made it up on the spot!

An LLM hallucinating is exactly this auto driver. It strings words together that sound incredibly correct and plausible, but are completely disconnected from reality.

Try It Yourself