Considering Responsible AI with GitHub Copilot
Kurt Dowswell · 2024
This chapter introduces responsible artificial intelligence (AI). It examines how GitHub Copilot implements Responsible AI. The chapter focuses on programming with AI Responsibly. Microsoft has established responsible AI principles across six categories: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. The idea behind the fairness principle is that “AI systems should treat all people fairly”. Microsoft believes that “All AI systems should perform reliably and safely”. Red teaming is the practice of adopting an adversarial approach to identify vulnerabilities in the reliability and safety of a software system. GitHub Copilot employs a vulnerability prevention system that inspects generated code suggestions for security vulnerabilities before returning the suggestion to user. In the development process, GitHub product teams partner with the Microsoft Office to Responsible AI to move the product through the established responsible AI processes.