PORTALE DELLA DIDATTICA

Ricerca CERCA
  KEYWORD

Methods for identifying code generated by artificial intelligence

keywords ARTIFICIAL INTELLIGENCE, COPYRIGHT, SOFTWARE DEVELOPMENT

Reference persons RENATO FERRERO, PAOLO GIACCONE, ENRICO MASALA

Research Groups DAUIN - GR-05 - ELECTRONIC CAD & RELIABILITY GROUP - CAD, DAUIN - GR-11 - INTERNET MEDIA GROUP - IMG, Telecommunication Networks Group

Thesis type RESEARCH AND DEVELOPMENT

Description The use of artificial intelligence (AI) for writing code is becoming increasingly widespread: AI provides additional tools and techniques to assist human programmers in the software development process. Using AI to write code can be legitimate and beneficial in many contexts, but it is important to carefully consider the legal, ethical, social, and economic implications. For example, using AI to generate code and present it as your own without giving credit to the origin could constitute plagiarism or violation of academic rules, especially in contexts such as exams or programming assignments.
Currently, there are no specific programs to determine whether code was written by an artificial intelligence. The thesis activity concerns the analysis of techniques to identify any clues that could suggest the origin of artificial intelligence:
- analysis of code characteristics: the code is examined to identify patterns, programming styles or structures that are typical of code generation by an artificial intelligence.
- language pattern detection: Some natural language analysis tools could be used to detect language patterns in code that could suggest origination from an artificial intelligence.
- use of plagiarism tools to detect similarities between the source code and code models generated by specific artificial intelligence tools.

Required skills programming skills, natural language processing, data analysis


Deadline 24/04/2025      PROPONI LA TUA CANDIDATURA