[QUOTE=Hill Walker;1273235]The first thing that should be recognised in any discussion about ‘Artificial Intelligence’ is that it is completely misnamed. Intelligence is one of those things that is usually easy to recognise but is almost impossible to define, like (for example) beauty it has intangible properties. It is therefore impossible to produce an artificial version of something which cannot be defined.
The second thing is that the term ‘Artificial Intelligence’ is being used to cover a range of things that are completely different and have no fundamental connection to each other.
For example ‘Computer Image/Vision Recognition’ is based on so-called ‘Neural Networks’ the design of which was an early attempt to produce a computer model of a very tiny bit of how brain tissue was thought to be constructed. They do work and are useful, BUT they have to be trained and it is almost impossible to be certain quite what they have been trained to do. For example self driving cars are trained to recognise people crossing the road but this can fail when the situation differs from the training, if for example two people carrying a long ladder cross the road the software will probably recognise the people but could fail the recognise that they are connected by the ladder.
On the other hand ChatGPT is based on a (language) syntax analysis programme which tries to work out what the user said, the results of this analysis then triggers a reply which usually is derived from searching a database of previous replies and/or searching the larger ‘database’ of a web search engine.
‘Artificial Intelligence’ covers both these examples (and many others) but the only thing that is common is the use of a computer.
The above is greatly simplified but I’m sure you’ll get the idea.
(Also note that Evolutionary Biologists are still undecided if ‘Intelligence’ is a Darwinian survival characteristic so it could all be a waste of time and effort.)
A well put together post but I am even more confused now,
