AI isn’t the only one that hallucinates

We often hear about AI models creating false information. We cringe at the lawyer who unknowingly submitted a legal document with made-up cases generated by ChatGPT.

A colleague asked the chatbot for a chart on the training costs of AI models and was surprised to find that the priciest one listed didn’t even exist (unless the AI has some insider knowledge). Just like humans, these models can be unreliable, as many users quickly discover. The real question is: are machines easier to correct than us? Perhaps it’s simpler to fix code than rewire the human brain.

Read the full article at: