Highly esteemed "Knowledge Exchange" collaborators and scholars, I am wondering if we can attempt a crystal clear analysis of literary texts with the assistance of artificial intelligence, and if it is to be done, how accurate will it be?
It depends on how you define " crystal clear analysis "
Some of the questions to answer are:
What is the level of automation that you are expecting? Is it full automation or partial? Is it to supplement human understanding or stand alone analysis?
What is the complexity of the literary work that will be analyzed? What background information(e.g. historical data) is needed to contextualize the work?
Other questions will come up as one tries to define the scope of the work.
It depends what you are analysing, but anything that can be defined and counted, e.g. an author's favoured vocabulary or narration of specific types of action could be extracted -- a lot of work on language and style has already been done and can help to settle disputes about authorship.
I suspect that this limited framework is the most likely domain for AI if you are looking for serious scientific insights.
On the other hand, the same AI capability provides a good basis for artificial literary art (and music). e.g. computer-generated fiction. I've written a novel, Whispering Crates, based on the concept of a fiction-writing application.
It depends on what "literary" means in the particular analysis. AI could be very useful in detecting a collaboration, by identifying characteristic language patterns of authors or usages that indicate later alterations, e.g. an 18th century word found in a 16th century text. AI could also provide statistics on various aspects of language or content, and could add in many other ways to the factual knowledge about a literary work. AI has already been used for this sort of work. However, when it comes to evaluating the aesthetic merits of the work and its artistic value, I feel that human criticism will always have the ultimate authority (though of course there will be disputes!) because only a human being can have the sensibility and experience to judge whether the artist has done his/her task well by reflecting real life, producing an effective fantasy, evoking appropriate reactions to an occasion etc.
This does not exclude the possibility of an AI system using analysis of classics and human culture to generate very convincing imitations of good writing or other artworks -- but such productions are likely to show their limitations sooner or later. The same is true of human authors who write good pastiche.
Eventually, AI may be capable of reproducing human consciousness, including the full range of reflexes, passions etc., in which case computers will be on an equal or even superior footing as critics.