As generative AI becomes increasingly integrated into software development, educators are reassessing how students should learn programming and how instructors should evaluate learning when AI can generate code.
A new report from the Association for Computing Machinery (ACM) highlights growing concern among computing educators about the influence of generative AI on student learning and assessment. The study underscores the need for new approaches to programming instruction.
Based on responses from more than 750 educators across 49 countries, the report provides one of the broadest views to date of how computer science instructors are responding to AI coding tools such as ChatGPT and GitHub Copilot.
The findings show both opportunities and challenges:
• 69% of educators believe the skills required to create software have changed due to generative AI.
• 87% identified increased dependency on technology as a concern, making it the most frequently cited challenge.
• 72% expressed concerns about cheating and plagiarism, while 53% cited misinformation.
• Nearly half pointed to a lack of best practice examples as a barrier to integrating AI into programming education.
• More than two‑thirds reported modifying assessment approaches in response to AI.
The report suggests that the conversation in computing education has shifted from whether AI tools should be allowed in programming courses to how they can be incorporated while maintaining core learning outcomes. Many instructors reported placing greater emphasis on code comprehension, program design, debugging, testing, and critical evaluation of AI‑generated outputs. Oral exams, code reviews, and project‑based assessments are increasingly being used to evaluate how students approach programming challenges.
Respondents also pointed to the lack of established best practices as a significant challenge. Many expressed interest in professional development focused on assessment design, classroom implementation, and responsible AI use, highlighting the need for additional resources as institutions adapt to AI‑assisted learning.
The task force concluded that while no single model has emerged, computing educators are actively experimenting with new teaching and assessment approaches to balance student learning, academic integrity, and workforce preparation in an era of AI‑assisted software development.
The ACM Education Advisory Committee’s Task Force on Generative AI and Programming Assessment conducted the global survey between May and October 2025. It received 763 responses from educators in 49 countries, with approximately 500 complete responses included in the primary analysis. The task force also reviewed instructor‑submitted examples, institutional approaches, and community‑contributed practices to inform both the final report and a broader ACM initiative to share emerging approaches with the computing education community.
Click here to read the original press release.
