By: Michael Flierl and Jane Hammons
The rapid growth of AI technologies has generated concerns and questions for both educators and students. Here are a few things that you should know about AI.
AI: What We Know
- Rapid Growth: AI systems are developing rapidly across wide sectors of education, science, culture and society. AI technologies are contributing to the development of new malaria vaccines, have passed the Uniform Bar exam and are becoming incorporated into military applications involving lethal force.
- Lack of Reliability: Current generative AI technologies are neither reliable nor trustworthy. As of July 2023, the state-of-the-art systems appear to be accurate, at best, 80% of the time on certain subjects. New phenomena are also emerging like “Drift” describing how ChatGPT-4 solved a math problem correctly 97.6% of the time in March 2023 but only 2.4% of the time in June 2023.
- The Black Box Problem: Current widely-available AI chatbots using neural networks are “black boxes,” computer scientists do not fully understand the millions or billions of calculations between inputs and outputs. Developing explainable AI (XAI) is an emerging research area.
- Detection: As of July 2023, AI detection software is easily fooled. Such systems are not currently reliable to detect the use of AI for assignments. Furthermore, accounting for demographics, effort and expertise, humans are poor at detecting AI-generated language.
- Educational Experiments: Universities like Harvard and non-profit tutoring services like Khan Academy are experimenting with using AI pedagogically. Some educators are embracing an open AI policy—focusing on students reflecting on their use of the technology.
How should you deal with AI technologies in your own teaching? Here are some action steps:
- Learn More: The U.S. Department of Education advocates for an educator-centered approach to AI technology. To get started learning more about AI and higher education, read the following report: Artificial Intelligence and the Future of Teaching and Learning.
- Experiment with AI Technologies: Input your own assignments into ChatGPT or Claude 2 and review the results. Think about the ways you could use AI technologies as learning tools, rather than only thinking about them as tools for cheating. Consider the strategies and examples on the AI: Considerations for Teaching and Learning guide in the Teaching and Learning Resource Center.
- Investigate AI Detection Software: As already stated, AI detection software is not accurate. Before making any decisions about if and how you will be using such software in your classes, do some additional research about how these tools work. A good place to get started is AI Detection Tools Are Really Easy to Fool.
- Be Transparent: The Committee on Academic Misconduct at Ohio State is already seeing cases of academic misconduct related to the improper use of generative AI technology. To help avoid such situations in your courses, clearly outline your policies about the use of AI in your course syllabus and in assignment instructions. Have a discussion with your students early in the semester about what is and what is not acceptable use of AI technologies in the course. Return to the discussion often throughout the semester. And remember that AI is much more than ChatGPT. If you have specific policies related to “homework help” sites such as Chegg or Course Hero, or tools such as Grammarly or paraphrasing generators, be sure those policies are clear for students.
- Think Beyond the Writing Assignment: Although much of the discussion around AI has centered on the use of AI as a writing tool, remember that AI can also generate artwork, spreadsheets and computer code. Just because you don’t have traditional writing assignments in your courses, doesn’t mean that AI won’t impact your course. Be sure to be clear about your policies for the use of AI in other situations.