Post by Purple Penguin PA

AI sounded confident… but it was confidently wrong. This week I had a useful reminder that, while AI can be brilliant, it’s definitely not perfect. I was asked ChatGPT to write multiple bios of around 165 words to fit in with a specific format of template I was producing. It had already created several with no issues - but one came back and looked noticeably longer, so I asked whether it was the correct length. It confidently told me: 👉 “the bio I’ve just provided is 158 words.” Except…it wasn’t. I pasted it into Word and the real count was 185 words! When I queried it, I initially got explanations about paragraph spacing and line wrapping (which don’t affect word count). After a bit more pushback, it finally accepted the error and confirmed it would avoid estimates next time. 💡 The takeaway? Trust the tool - but always verify the output. AI is fantastic for speed, structure, and creativity. But it can: • Miscount • Misinterpret • Miss the brief • Sound certain while being wrong So no AI isn’t going to replace your job… BUT if you start handing over unchecked AI output… it might! Quality still matters. Accuracy still matters. Human judgement still matters. AI is an accelerator - not a substitute for accountability. Have you experienced anything similar?