Problem Solving

We listened to a sceptical expert about 'large language models'. He said that they will not be the basis of general intelligence, because they cannot take the knowledge they have stored and apply it to new scenarios that are not closely analogous to the ones they have learned. People like him keep creating examples of this shortcoming; the subsequent release of the software can solve the example problem, but that is because the model has absorbed the problem and the solution in the published work about it, not because it has worked it out for itself

Our grandsons had a wonderful time at Glastonbury - a lot of it together. Imagine how much learning can happen in an environment like that. We caught up with one of them today to hear his stories. His mum said that one day he spent a long time, repeatedly taking a cereal bar from its box, via a small hole in the top, and posting it back. As soon as he got home, he took one of his toys, pulled out a wooden post and then put it back - something he had not mastered before he left. Grandson: 1; ChatGPT: 0

He loves bread, but bread crust is a problem when you don't have many teeth and all of them are incisors (a scenario I'm apprehensive about myself). He hasn't given up trying to solve the problem, but looks a little confused when he has to give up and spit it out. He has no way to know that nature will solve the problem for him 

His mum bought cakes from the new baker in town, as a treat for the grandparents (very good). Also a baguette for the family supper. If I cooked my bread until the crust was as 'browned' as almost all the bread here, I know MrsM would have a word - she shares some of her grandson's views on tough crusts. So I don't; problem solved