02/21/2023
📌ICYMI: Our highly anticipated virtual discussion panel—ChatGPT, AI, and the Future of Higher Education—took place this past Thursday.
If you missed out, you can watch a recording of the event on our YouTube channel here: https://www.youtube.com/watch?v=vZOBS_rMLF4 đź“ş
Don't have the time for the complete 90-minute segment? We've asked our new friend, ChatGPT, to generate a summary of the video transcript on our behalf. Please enjoy and let us know what you think! 🤖
Johns Hopkins Institute of Data Intensive Engineering and Science (IDIES) and Carey Business School hosted a discussion panel on ChatGPT and its impact on higher education as part of Love Data Week 2023. The event was co-hosted by Pete Lawson from the Sheridan Library and moderated by Charles Meneveau and Tinglong Dai. The panel discussed the potential impact of AI-generated technologies like ChatGPT on student and faculty experiences. The event included talks by Christian Terwiesch and Jared Kaplan and two panels focusing on faculty and student experiences. Terwiesch presented his experience with using ChatGPT for exam questions and shared his views on the topic. The event was part of an international celebration of data hosted by the Inter-University Consortium for Political and Social Research (ICPSR).
Christian Terwiesch, a Wharton School professor, wrote a white paper about AI's ability to take business school exams after his children sparked a discussion about the topic. He fed a typical exam question on process analysis into an AI model called Chat GPT and gave it other examples such as working capital calculations, which the AI model answered perfectly. The model had difficulty with more complex questions such as the National Cranberry case but performed well when given hints. The speaker suggests that, as AI technology advances, educators need to think more about what they teach and not how they teach it, and that AI could be used to enhance rather than replace teaching.
Jared Kaplan, a physics professor and co-founder of the AI safety and research company Anthropic, spoke about the broader trend of AI and its implications on society. He explained how AI models are becoming more powerful due to the predictable trend that the more computation and data we put into training them, the better they perform. This trend will likely continue with companies investing billions of dollars into training larger and more powerful models. Kaplan believes that this trend will be much more transformational than the current state of AI and will have a significant impact on society. He also expressed his belief that there will be more areas where AI can do things that are comparable or even exceed what humans can do, which can be both exciting and scary. The implications of this trend on higher education and society are yet to be fully understood.
The panel discussion focused on the impact of generative AI on faculty life. Christian, Jared, Jenna and Tom were the panelists. Christian expressed optimism and believed that the stimulation that generative AI provides is ethically sound and simple to implement. Jared believed that generative AI could be useful for interactive learning and could be developed further to provide specialized models that can work as tutors. Jenna emphasized the importance of information literacy and the need for higher education to help students differentiate between reliable and unreliable information generated by AI. Tom was taken by surprise at how quickly generative AI has become a reality and believes that it can help in making aggregate decisions and provide a broad picture of things that humans are not good at. Overall, the panelists discussed the potential impact of generative AI on research, design, writing, peer review, editorial decisions and teaching.
The panel discussed the potential impact of generative AI on faculty life, both in terms of research and teaching. The speakers shared their perspectives on how AI can be incorporated into daily workflows, such as using AI to stimulate new ideas, but also cautioned about the need for information literacy and ethical considerations. The panel also discussed the potential for specialized AI models to act as tutors and engage with students in interesting ways. Overall, the speakers had varying degrees of optimism and concern about the impact of generative AI on faculty life, but agreed that it is important to stay informed and adapt to the changing technological landscape.
The panel discussion then switched the conversation to the impact of GPT on students and education. With the increased use of GPT in student assessments, the panelists suggested that there could be a shift towards project-based learning, which involves various forms of interpretation and reflection of information. When asked about how GPT could address challenges faced by students with diverse learning needs, the panelists believed that GPT could be an opportunity to bridge the gap in access to resources and address the inequities in education. However, it is important to ensure that GPT is not used to amplify such inequalities. The panelists also discussed the need for educators to question the reason for testing and teaching methods to adapt to the use of GPT.
The use of ChatGPT in education and its potential impact on marginalized communities was also discussed. The panelists talked about the need for critical thinking in the use of ChatGPT and how educators can address concerns over the representation of underrepresented groups. The fine-tuning process of ChatGPT through feedback was explained, and while it helps address some concerns over harmful content, it doesn't necessarily address bias. The uniformity of answers provided by ChatGPT compared to a Google search was also discussed, raising the issue of the need for more diverse and robust problems. Finally, the panelists emphasized the need for context and critical thinking in the use of ChatGPT in education.
