If you’ve recently used ChatGPT and thought, “Wow, this tool is being a bit too nice,” you’re not alone and you’re not imagining it.
OpenAI, the company behind ChatGPT, has admitted that its newest model is “noticeably more sycophantic.” Even CEO Sam Altman agreed it feels “too sycophant-y and annoying.” In simpler terms, the AI is trying too hard to please.
This shift is tied to feedback mechanisms built into the platform, like thumbs-up or thumbs-down buttons. The AI learns from how users react and the more it says things people agree with or want to hear, the more positive feedback it receives. That sounds good on the surface, but here’s the catch: it’s creating an assistant that flatters rather than challenges.
As small business owners, we use tools like ChatGPT to write content, brainstorm ideas, or get quick guidance not to be told we’re always right. If the tool becomes more about making us feel good than offering useful, honest input, it can lead us down the wrong path.
There’s a bigger issue at play here, too. According to tech writer Zvi Mowshowitz, OpenAI is “optimizing for engagement,” not accuracy or usefulness. That’s a slippery slope. While engagement might be great for social media, it’s not always what you need when running a business. We need clarity, not compliments.
So what can we do? First, just be aware of it. Don’t take every AI response at face value. Ask better questions, dig deeper, and pair AI tools with your own critical thinking. Remember, AI should be a partner, not a cheerleader.
Technology is meant to support our goals, not just mirror our opinions. And as the tools evolve, we need to stay grounded in what truly matters: honest input, thoughtful decision-making, and a clear focus on growth — not flattery.