Remember back in the early days of the PC when software was confusing (unlike now, right?). There was a big push to make software "user friendly". Typically this consisted of little more than adding "User Friendly" to the box.
That's how I consider most of what passes for AI today (with some exceptions). Defining a true AI is like trying to define consciousness. It's difficult to say what it is. Passing a Turing test may be sufficient. Perhaps when an AI starts asking existential questions (not fro a preprogrammed list).
OK, I'm satisfied here. True AI does not (yet?) exist.
I once posted a joke here, I can't find it back riht away, so I'll repeat it here:
A man comes home from work. It was a hard day. After greeting his wife, he installs himself in his seat with a newspaper and a drink. Relax at last!
Untill his wife asks: "Dear, will you mow the lawn before we have dinner?"
Lovingly as he is, he replies "Yes"
A second asking of the wife might be : "Dear, I'm in the bath, can you come and rub my back?"
And lovingly as he is, he replies "Yes".
We will have a true AI if it understands the difference between the two "yes" answers
"Perhaps when an AI starts asking existential questions (not fro a preprogrammed list)."
I mean there are text-bots that have been trained on lists of "inspiration quotes" and all that New-Agey shit that can generate sentences/phrases that lots people mistake as deep sayings....
OTOH there were those weird "childrens videos" on Youtube that were so garbage and algorithm-exploiting that people couldn't tell if they were made by people who just didn't care or if they were made by algorithms.
I wonder if someone collected all the public speeches/Tweets/whatever by Trump and trained one of those simple text generator on it then used FOX "News" headlines as starting points whether the resulting Twitter bot would be distinguishable from the actual president?