Student Builds Local Test Witness to Retain Ownership of AI-Assisted Lab Work
A computer science and AI student in Halifax faced an ethical dilemma while completing a 40-line name-normalization lab assignment last Tuesday, questioning whether submitting AI-generated code truly represented their own understanding. The student used MonkeyCode, a free open-source tool, to build a local 'witness' system that sends only a constrained demo string to a remote model while keeping all fixture data and expected outputs stored offline. The setup logs cryptographic hashes and pass/fail results locally, ensuring course roster data never leaves the laptop and the remote model cannot see the answer key. By isolating the remote draft call behind a single swappable function, the student could evaluate AI suggestions without surrendering grading authority to an external machine. The project frames AI assistance as a photocopier, not a grader, arguing that locally held test fixtures remain the only trustworthy evidence of genuine comprehension.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in