I’ve had a powerful response to a piece I wrote on my Substack about an experience with AI. I received lots of private comments and messages, some saying that students and teachers should all read it urgently so that they can see AI-generated answers for what they are: highly plausible, confident and brilliant but with absolutely NO honesty or reliability. Telling the truth is not the aim of the AI response. The aim is to answer your question in a way that you will like.
This is what happened when I tested it out (using Co-Pilot, Microsoft’s AI tool) and it failed, brutally. The apologies I received were nauseating and even terrifying in their clinical dishonesty.
AI can be a most wonderful tool, with immensely positive applications. But we need our far more brilliant human brain to use it safely.
For schools, the real challenge is how to nurture and develop the brilliance of the human brain. After all, use it or lose it. THAT is what I’m most worried about.
One Response
Fascinating but very disturbing. Out of curiosity I asked Co-pilot about my own work. The response was very similar. No, I do not trust AI!