Definition
Generative AI refers to a type of artificial intelligence technology that is designed to create new and original content, such as images, text, music, or other forms of data, by learning patterns and characteristics from existing examples. It can mimic and produce outputs that resemble human-made creations. Here's a video that explains Generative AI.
Capabilities
Generative AI capabilities are constantly improving. Below are just some of the things that Generative AI engines can assist with:
- Natural language understanding and generation
- Answering questions
- Language translation
- Summarize and synthesize information
- Computer programming help
- Mathematics and calculations
- Conversational partner
- Creative writing
- Image, video, and music generation
Limitations
- Some generative AI tools struggle to consistently incorporate legitimate research into text. Although certain tools may generate text accompanied by a reference list, the references often deviate from the generated content or consist of fabricated citations that blend factual publication details from various origins. More advanced iterations, such as GPT-4, exhibit enhanced proficiency in seamlessly integrating research materials.
- Inability to provide timely responses or address inquiries pertaining to recent events. ChatGPT's training dataset extends up until 2021, and ongoing efforts are in place to enhance its database.
- Potential to generate responses that could be inaccurate, excessively simplified, unsophisticated, or influenced by bias when presented with questions or prompts.
- Generative AI may not always be entirely accurate and can potentially produce false or misleading information. Its responses are generated based on patterns in the data it was trained on and may not reflect the most up-to-date or reliable information. Users should exercise caution and critical thinking when interpreting and relying on its output, especially when seeking factual or authoritative information. Cross-referencing and verifying information from trusted sources remains essential to ensure accuracy and reliability.
Educational Uses
- Research Assistant
- Language learning
- Writing support
- Concept clarification
- Creative writing workshops
- Coding and programming assistance
- Simulation and role playing
- Ethical and philosophical dialogue
- Historical and cultural conversations
- Business and Marketing scenarios
AI and Privacy
ChatGPT and other Generative AI tools may share personal information collected from the user with third parties, including vendors. They may also use the information provided to train the system which may then include personal information. Talk to your students about responsible use of the systems that does not include personal information and encourage them to read the Terms of Agreement and the Privacy Policy prior to signing up for any tool.
Detection
A wide variety of tools that claim to recognize AI generated text have appeared. While their accuracy is constantly evolving significant challenges remain and many of the tools have a high rate of false positives. While the accuracy of the systems can improve, there are also concerns that Generative AI tools will continue to adapt to evade detection. Care should be taken to avoid viewing them as complete solutions. As such, the Office of Online Learning will not support or endorse any of the products currently available. We will continue to monitor the available tools and will update the campus as changes occur.
Academic Integrity
- In your syllabus, inform students whether they are allowed to use Generative AI programs in your class and if yes how and for what purposes.
- Talk to your students about academic integrity and why it matters.
- Encourage students to verify that the content generated by the model is a genuine reflection of their knowledge and understanding. Teach students to critically evaluate the information provided by Generative AI.
- If you plan on allowing Generative AI use in the classroom, here is some information on how to cite it appropriately: APA guidance; other citation formats.
- If possible, reduce the number of high stakes summative assessments and include more frequent, low stakes, more formative assessments.
- If possible, create assignments that are harder to complete with AI (e.g., personal reflections or assignments that include more recent events that are not in the AI database).
- If teaching in person, consider flipping the class where students read materials and view lectures at home and then apply, demonstrate, and perform in class.
Additional Resources
- AI in Higher Education Metasite
- Cotton, D., Cotton, P., & Shipway, J.R. (2023, January 10). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. EdArXiv Preprints.