AI Still Gets Confused About Basic Rules

You've probably heard a lot about how clever Artificial Intelligence (AI) is getting. It can write emails, summarise documents, and even create images. But here's a recent example that shows even the best AI models still have some pretty big quirks – especially when it comes to understanding and applying specific rules.

A law professor was using Claude, a well-known AI, to ask a question about the hiring process for law clerks in the US. It was a very specific query about when someone could contact a judge to recommend a student. Claude, in its response, actually told the professor that judges wouldn't accept applications or recommendations directly. The problem? That's not quite right. While there are strict rules, the AI seemed to misunderstand the nuances of the process, giving an incomplete and potentially misleading answer.

This isn't just a one-off. It highlights a common challenge with today's large language models (LLMs) like Claude. They're brilliant at sounding confident and articulate, but sometimes that confidence masks an outright error, or what AI experts call a "hallucination". It's like talking to someone who's read a lot of books but hasn't quite grasped the real-world application of the information.

For Aussie small business owners, parents or anyone using AI, this is a really important lesson. While these tools can be incredibly helpful for drafting ideas, summarising text, or getting general information, you can't always trust them blindly. Always double-check any critical information, especially if it relates to laws, regulations, financial advice, or anything that has serious consequences if it's wrong. Think of AI as a very clever, but sometimes a bit scatterbrained, intern – helpful for the first draft, but always needs a thorough review.