Coming up with arguments to defend uniqueness of a thesis document

I have examples of PDF documents that contain objections raised against the novelty of research. A typical document will contain objections such as “In your thesis, you say that the way you have performed a Machine Learning technique is new. However, your technique is very similar to the technique proposed by XYZ researcher. The XYZ researcher uses a pooling of layer technique and your thesis performs an operation that is very similar to pooling”. The document will also contain the response of the researcher. Example response can be, “Even though I performa similar pooling technique, the way I derive the number of layers is very different from the XYZ researcher”.

I am planning to build a system that understands these thousands of examples and prepares arguments when a new objection is raised. The domain of the research papers will always be fixed (meaning all the documents belong to similar technical areas). How best I can leverage OpenAI? Thank you.