By this definition a how to book is intelligent.
Classic programming is just a list of instructions (steps to take) that the computer follows exactly as written by the programmer. It can appear complex or magical, but we fully understand exactly what will happen in every instance. Nothing unexpected or new ever happens.
AI as it is today mixes things up just a bit. We allow the computer to program itself by training it on lots of data while it builds up a neural network (think of it as a fancy decision tree) that it can then use to try and guess if the next data matches the training data. This is great, amazing at times, but in a lot of ways it is automated programming (auto classification) and not really intelligent in any sense of the word.
The magic happens when computers can become intuitive and make a leap. Say we show it lots of apples and lots of tennis balls and tell it one is an apple and one is a tennis ball. Then we tell it an apple is a fruit and a tennis ball is a ball. Can we then show it a soccer ball or an orange and have it intuit that they are fruit or balls. This is the challenge, and we really don’t know how to get there yet, partly because we don’t know how we do it ourselves.
I wish this were true! The problem with Linux is that it is constantly changing. I have been using it for 30 years and have built my own embedded distros from scratch. Yet every time I turn around there moving this setup file to another directory or changing out that language for a slightly incompatible newer version. Trying to configure and maintain a box is a constant battle.
Windows is the polar opposite. The ui may have some annoying changes but under the hood it is frustratingly stable, often remaining unchanged for 20+ years (even the bugs live forever). Users crave simplicity and consistency. It is something Linux still needs to figure out.