The rapid adoption of Artificial Intelligence (AI) and Large Language Models (LLMs) such as ChatGPT in higher education has led to a paradigm shift in programming instruction and assessment. Traditionally, take-home coding assignments and standardized exercises served as a robust means to assess student competency in programming. However, with LLMs now capable of generating, debugging, and optimizing code with minimal human input, these methods risk becoming obsolete, compromising the authenticity and effectiveness of skill assessment in educational contexts.

This study investigates the constraints of existing assessment models in the context of AI-powered automation and investigates alternative approaches to guarantee skill assessment and knowledge retention. Proposed solutions include live coding interviews, which assess real-time problem-solving and adaptability; collaborative coding challenges, including peer interaction and collective problem-solving; and project-based assessments evaluating the student's ability to integrate knowledge over extended, multifaceted projects. Additionally, we discuss the use of oral examinations that require students to articulate their understanding of programming logic and AI-generated solutions.

Ethical and practical considerations are also addressed, particularly regarding the integration of AI tools in educational settings. Potential challenges, such as overreliance on AI assistance and concerns about academic integrity. We argue that embracing AI within the assessment process, rather than excluding it, provides an opportunity to improve critical thinking and adaptability in students.

The main objective of this work is to provide a framework to help educators transition to more resilient AI-adaptive models of skill evaluation. This approach aims to preserve the core objectives of programming education, ensuring that students develop practical, adaptable, and critical skills in an AI-influenced educational landscape.

Keywords: AI in education, LLMs, Critical Thinking, Skill Evaluation.