WIP: Focusing on Programmer Literacy in the Time of AI-Aided Code Generation

This work-in-progress, innovative-practice, full paper describes an approach to the training of computer scientists and engineers focused on the reading of programs generated by AI-based Large Language Models (LLM). In contrast to the practice of Literate Programming, in which the coder crafts a solution which reads as English, the proposed approach treats an existing programming as a “shorthand”, which, when articulated, is explanatory. LLMs amount to enhanced search, having digested numerous examples with attention to context in order to provide a most “desired” result for a given prompt. How well the LLM generates a solution for a particular language varies substantially. The paper gives numerous LLM-generated code examples of how the proper reading of programming statements conveys their function and requires little or no commentary. The Sieve of Eratosthenes will be used to contrast the LLM-generated code against code which adheres to the language's idiom.