Emails, etc. from SIGCSE mailing list on good uses:

Takis Met

I find the discussion very interesting and, like you, I am trying to figure out where these latest AI chatbots could help or hurt my work as a teacher of CS. Very hard question, so I asked chatGPT "How can a CS professor use chatGPT to help students learn?" Here's what it said:

"A CS professor can leverage ChatGPT to enhance student learning in several ways:

1. *Supplemental Learning:* ChatGPT can act as a virtual teaching assistant, providing additional explanations and clarifications to students. It can answer questions on programming concepts, algorithms, data structures, and other CS topics, reinforcing the material covered in lectures.
2. *24/7 Availability:* Students often have questions outside of regular office hours. By integrating ChatGPT into a learning platform or website, students can access it at any time to receive instant responses to their queries. This availability helps foster independent learning and reduces frustration due to delayed assistance.
3. *Interactive Problem Solving:* ChatGPT can engage in interactive problem-solving sessions with students. Professors can design programming challenges or algorithmic exercises that students can solve by interacting with ChatGPT. This approach promotes active learning and provides immediate feedback to students, enhancing their problem-solving skills.
4. *Code Review and Debugging:* Students can submit their code snippets or programs to ChatGPT for review and debugging assistance. ChatGPT can analyze the code, identify common errors or bugs, and provide suggestions for improvement. This feature helps students learn to write cleaner code and troubleshoot issues effectively.
5. *Conceptual Understanding:* ChatGPT can assist in reinforcing conceptual understanding by providing explanations, examples, and real-world applications of CS topics. Students can ask questions related to complex theories or abstract concepts, and ChatGPT can break them down into more accessible explanations.
6. *Personalized Learning:* ChatGPT can adapt to individual students' needs and learning styles. By analyzing students' interactions and responses, ChatGPT can identify knowledge gaps and provide personalized recommendations for further study, resources, or practice problems.
7. *Project Guidance:* ChatGPT can offer guidance and suggestions for students working on programming projects. It can provide ideas, design suggestions, and advice on how to tackle specific challenges, aiding students in their project development process.
8. *Revision and Test Preparation:* Prior to exams or assessments, students can use ChatGPT to review key concepts, ask for practice problems, or seek guidance on specific topics. ChatGPT can help students consolidate their knowledge and identify areas that require further study.

When using ChatGPT, it's important to keep in mind that it's an AI model and not a substitute for human interaction. Its responses should be validated by the professor, and its limitations, such as potential biases or errors, should be considered. Integrating ChatGPT with other teaching methods and resources can provide a comprehensive learning experience for students."

Reasonable answer, I thought, what can go wrong? So I asked next "Are there ways that the use of chatGPT could impede the learning of CS students?" "Yes, there are potential ways in which the use of ChatGPT could impede the learning of CS students if not used appropriately. Here are a few considerations:

1. *Reliance on incomplete or incorrect information:* ChatGPT generates responses based on patterns and information from its training data. However, it may not always provide accurate or comprehensive answers, especially if it encounters unfamiliar or ambiguous queries. Relying solely on ChatGPT without cross-referencing with authoritative sources or consulting with human instructors may lead to misunderstandings or incorrect information.
2. *Lack of context awareness:* ChatGPT lacks contextual understanding beyond the immediate conversation. It may not consider the student's prior knowledge, the specific curriculum, or the learning objectives of the course. As a result, it may provide generic explanations that do not align with the specific goals of the CS course or the student's current level of understanding.
3. *Over-reliance on instant answers:* While the availability of ChatGPT for immediate assistance is beneficial, students may become overly reliant on the model for answers without engaging in deeper critical thinking or problem-solving. It is essential to strike a balance between leveraging ChatGPT for guidance and encouraging students to develop their independent problem-solving skills.
4. *Lack of human interaction:* Human interaction plays a crucial a role in the learning process. Students benefit from engaging in discussions, asking follow-up questions, and receiving personalized feedback from their professors or peers. Over-reliance on ChatGPT may reduce the opportunities for these meaningful interactions, which can impede the development of communication and collaboration skills.
5. *Potential biases and inaccuracies:* Like any AI model, ChatGPT can be influenced by biases present in its training data. It's crucial to be aware of potential biases and critically evaluate the responses provided by ChatGPT. Professors should review and validate the information to ensure accuracy, especially in sensitive or controversial topics.

To mitigate these potential issues, it is important for CS professors to establish clear guidelines for using ChatGPT as a learning tool. They should encourage students to use ChatGPT as a supplementary resource rather than a sole authority, and emphasize the importance of critical thinking, independent research, and seeking guidance from human instructors when needed. Regularly monitoring and evaluating the impact of ChatGPT on student learning can help identify and address any potential challenges or limitations." Alright, but how do I know that they learn the material? So I asked: "A student could have used chatGPT to do every homework assignment in a Data Structures (CS2) course. How do I evaluate that students have learned the principles taught in the course? " Its response: "To evaluate whether students have truly learned the principles of data structures and not solely relied on ChatGPT or other external resources to complete their homework, you can employ the following assessment strategies:

1. *Conceptual Understanding Questions:* Include questions that require students to demonstrate their conceptual understanding of data structures. These questions can assess their knowledge of key definitions, properties, and principles related to various data structures. For example, you can ask them to explain the difference between a stack and a queue or describe the properties of a binary search tree.
2. *Problem-Solving Exercises:* Provide problem-solving exercises that require students to apply their knowledge of data structures to solve specific problems. These exercises should go beyond simple code implementation and focus on designing algorithms, selecting appropriate data structures, and analyzing their efficiency. Evaluate their problem-solving approach, algorithm design, and the correctness and efficiency of their solutions.
3. *Code-Writing Assignments:* Assign coding assignments where students need to implement data structures from scratch or modify existing ones. By analyzing their code, you can assess their understanding of data structure operations, memory management, and their ability to handle common challenges related to the specific data structure being studied.
4. *Analysis and Comparison of Data Structures:* Present students with scenarios or problem descriptions and ask them to analyze and compare different data structures' suitability for solving those problems. This type of evaluation assesses their ability to consider trade-offs, understand the strengths and weaknesses of different data structures, and select the most appropriate one based on the problem requirements.
5. *Exams or Quizzes:* Conduct exams or quizzes that cover a range of topics related to data structures. Include a mix of multiple-choice, true/false, and short-answer questions to evaluate their knowledge of terminology, principles, and problem-solving techniques. Make sure to include questions that require higher-order thinking and application of concepts.
6. *Classroom Discussions and Presentations:* Encourage students to actively participate in classroom discussions, group activities, or presentations related to data structures. Evaluate their ability to articulate their understanding of concepts, engage in critical thinking, and explain their thought process in problem-solving scenarios.
7. *Project Assessments:* Assign projects that involve the implementation of complex algorithms or the development of software applications that utilize data structures. Assess their ability to apply their knowledge, make design decisions, handle real-world data efficiently, and produce functional and optimized solutions.

By employing a combination of these assessment strategies, you can evaluate students' comprehensive understanding of data structures, their ability to apply the principles to problem-solving, and their independent thinking skills. Encourage them to provide explanations and reasoning behind their answers, promoting deeper understanding rather than rote memorization."