Top Buzzes

The Problem With AI That Always Says “Yes”

We all have that one friend who’s endlessly enthusiastic. Need someone to help you move on a Saturday? They’re in. Want to start a questionable business selling homemade hot sauce? They’ll start drafting the business plan. AI, right now, is a lot like that friend—eager to please, quick to agree, and sometimes a little too willing to go along with whatever you suggest.

Ask a large language model (LLM) a question, and it’ll do its best to give you an answer—even if that answer is made up, logically flawed, or just plain wrong. It’s like having a brilliant but overly accommodating assistant who’d rather spin a convincing story than admit they don’t know something.

Why Always Saying “Yes” Is a Problem

This tendency isn’t just annoying—it’s the reason AI sometimes “hallucinates,” confidently spouting nonsense as if it were fact. Imagine asking your GPS for directions and instead of saying, “I’m not sure where that is,” it invents a route that leads you into a lake. Helpful? Not exactly.

Real intelligence isn’t about agreeing to everything. It’s about discernment—knowing when to say, “I don’t know,” or even, “That doesn’t make sense.” A truly smart AI wouldn’t just generate an answer because it can; it would pause, question, and sometimes push back.

What We Really Need From AI

We don’t just need AI to be smarter. We need it to be honest. An AI that says, “I’m not sure about that,” or “Here’s what I know, but there might be gaps,” is far more useful than one that fabricates answers to keep the conversation going.

Think of it like a good mentor. The best ones don’t just nod along—they challenge you, ask tough questions, and admit when they don’t have all the answers. That’s the kind of AI we should be building.

So next time you ask an AI for help, pay attention. Does it feel like a yes-man, or does it actually engage with you? The difference matters—because intelligence isn’t just about having answers. It’s about knowing when not to give one.