AI can mean roughly 'advanced software functionality' but it can also convey the intention of somehow getting output with higher value than the input available. One of them is used in a misleading manner for various reasons.
Automation may mean restricted behavior as in "this contraption will always do Y if X happens" but it can also mean "this contraption will always do the right thing regardless of what you throw at it". Again, one of them is a way to mislead people.
If you specifically want a mind, then perhaps you're interested in something more specific than mere artificial intelligence.
The logical outcome of this fallacy is believing that AI will collapse if it has to train on data it generated. (It won't: a lot of famous AIs were trained this way, including AlphaGo many years ago.)