Beyond the Hype: A Comprehensive Review of Current Trends in Generative AI Research, Teaching Practices, and Tools

Generative AI Tools for Code Support and Learning Enhancement

Context: Used in various computing courses, including introductory programming, software engineering, and artificial intelligence courses targeting both undergraduate students and instructors. The AI tools provide assistance in coding, debugging, code understanding, and learning support.

Implementation: Instructors integrate generative AI tools (such as ChatGPT, GitHub Copilot, and RAG tools) into their courses to provide coding support, feedback, and context-aware assistance. These tools are utilized for generating code, resolving compiler errors, and enhancing students' understanding of programming concepts and project outcomes.

Outcomes: Positive outcomes include improved student engagement, higher quality work, better understanding of programming concepts, increased success in error resolution, and overall enhanced project outcomes. Students also experience immediate feedback and a reduction in the need for office hours.

Challenges: Challenges include concerns about academic dishonesty, over-reliance on AI tools, potential bias in AI outputs, and the risk of reduced critical thinking and problem-solving skills if students do not engage deeply with the learning processes.

Retrieval-Augmented Generation (RAG) Tools for Context-Specific Learning

Context: Applied across various disciplines, including business and economics, to provide context-specific information that supports learning.

Implementation: Tools are customized to deliver relevant content and enhance learning interactions by adapting materials to specific course requirements.

Outcomes: The implementation leads to enhanced student engagement and a better understanding of course materials, allowing students to gain insights into their learning interactions.

Challenges: Challenges include the need for customization of tools for specific courses and ensuring pedagogical effectiveness.

Ethical/Policy Barrier

Concerns around academic integrity, plagiarism, and the potential for students to misuse GenAI tools.

Proposed Solutions: Implementing clear guidelines on acceptable use of AI tools in coursework, and requiring students to disclose AI use and provide documentation of their prompts and interactions with GenAI.

Technical Barrier

Difficulty in designing GenAI tools that provide useful and contextually relevant feedback to students.

Proposed Solutions: Developing customized tools with instructor-provided guardrails to improve the effectiveness of AI feedback.

Resource Barrier

Lack of time and skills among educators to effectively integrate GenAI into courses.

Proposed Solutions: Professional development and training for educators on GenAI tools.

Student Ability Barrier

Variability in students' skills and motivation to effectively use GenAI tools, which may lead to frustration.

Proposed Solutions: Providing training sessions on how to use AI tools effectively and targeted teaching strategies to enhance student competencies in using GenAI.

Skepticism Barrier

Doubts among educators regarding the usefulness of GenAI in teaching.

Proposed Solutions: Providing evidence from successful implementations and studies showing GenAI's benefits.

Institutional Barrier

Lack of institutional support and unresolved legal issues surrounding the use of GenAI.

Proposed Solutions: Creating guidelines and policies for the ethical use of GenAI in education.

Integration Barrier

Challenges in integrating AI tools into existing curricula and ensuring alignment with learning objectives.

Proposed Solutions: Collaboration between educators to design curricula that incorporate AI tools seamlessly.

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai