My children's generation may not be taught how to use AI
My children's generation may not be taught how to use AI. They may be taught how to catch it lying.
AP reported today that many US schools are opening the new term with AI lessons built around what AI gets wrong. In one teacher-training demo, an AI-generated world map spelled Mali as "Mail", Egypt as "Sopth", and labelled Libya, simply, "Africa".
The room gasped.
My first reaction was that it's funny. My second was that the map was chosen on purpose. The lesson doesn't open with what AI can do. It opens with where it fails.
I think that order is right.
I use AI agents at work every day, and the part I spend most time on isn't asking them to do things. It's reading what they produced and asking: is that true? where did that number come from? it sounds certain, but does it know what it doesn't know?
I use AI agents at work every day, and the part I spend most time on isn't asking them to do things.
That move is the same one our teachers taught us about newspapers: which paper printed this, and where does it stand? The difference is that the thing answering now is unfailingly polite, unfailingly confident, and carries no masthead.
One honest caveat: this is a US story. Whether English schools open this term the same way, I don't know. But I don't need the curriculum to move first, an eight-year-old is already extremely good at spotting mistakes. Mine delight in catching adults out on spelling. It's the same game with a new opponent.
And a child who can catch it isn't frightened of it. The frightened one is the child who assumes it's always right.
Do you play "spot the mistake" with AI at home yet, or is that the school's job?
