Please keep in mind that I am not an expert when it comes to AI/machine learning/Natural Language Processing and have just dabbled in those a few years ago. Also, many of these terms can be confusing because of all the marketing around them.

But, in short, what I meant was that autotldr is not based on a neural network/deep learning, but "basic" language processing algorithms with simpler statistical models attached to them, if at all.

What people generally mean with AI is some kind of deep learning which is helpful for more complicated tasks like "learning" about a topic and "understanding" things, but that is not really necessary for summarizing text since language has (mostly) fixed rules. So while there is certainly some overlap in the process, simple language processing does not need "AI" - you'd call it just machine learning, of which AI is generally a subset.

You can read a little more about the sunmarization above here:

https://miso-belica.github.io/sumy/summarizators.html