ChatGPT is here to stay: Generative Language Models and Chatbots in Education
The use of ChatGPT in today's classrooms and lecture halls has raised both anticipation and apprehension. In this interview, we sat down with experts Gergana Vladova, Diana Knodel, and Sofia Schöbel to explore the potential, as well as the challenges, that generative artificial intelligence brings to the world of education.
What can students learn from generative language models? What are the educational benefits?
Sofia Schöbel: Generative AI, such as ChatGPT, operates within a conversational framework. So particularly in the context of self-directed learning, engaging with a chatbot can be beneficial. Chatbots assist learners in gradually delving deeper into content through interactive conversations. This approach is often more engaging and eases the learning process. They can also be customized to adapt flexibly to a learner's individual pace and progress. ChatGPT serves as an illustrative example for students to comprehend the workings of generative AI, consequently shedding light on the role of data in artificial intelligence.
Gergana Vladova: The use of generative language models makes it possible, for the first time, to have a conversation with a technical entity that feels amazingly real. Very much in the spirit of Joseph Weizenbaum and his ELIZA – the first chatbot developed back in 1966 – which was supposed to act as a real conversation partner with the help of various scripts. It amazingly managed to do so, despite the conversations being very limited. Even back then, Weizenbaum pointed out that such developments should be viewed critically. And I hope that this – especially currently in the form of ChatGPT – will find its way into the classrooms in such a way that people learn to question the answers, to reflect and to use AI in a playful way, and to promote their own creativity and understanding of the technology.
Diana Knodel: Generative language models, such as ChatGPT, offer students a wide range of learning opportunities. By using AI assistants based on the technology, students can use personalized and adaptive learning environments. AI can also support creative writing and language expression by providing inspiration and valuable feedback on texts. It also provides easy access to extensive knowledge and can serve as an effective tool for research and information retrieval.
What risks are associated with using these tools in education, and how must learning material and methods adapt to address them?
Diana Knodel: Using generative language models in education can pose a risk where students might become overly reliant on AI assistance, potentially sidelining their own critical thinking and problem-solving skills. To address this concern, it's essential to view AI as a helpful tool that supports, rather than replaces, the learning process. When designing learning materials and approaches, the goal should be to encourage critical thinking, problem-solving abilities, and the capacity to handle uncertainty.
Sofia Schöbel: ChatGPT is trained by humans, and it learns from us. So, everything ChatGPT generates as answers is basically knowledge that hasn't been double-checked. That's why it's crucial to approach the information it calculates with a critical mindset and not just accept it blindly. In education, we should put more emphasis on teaching analytical thinking, problem-solving, and encouraging a healthy dose of skepticism. It's also important to help students understand how generative artificial intelligence like ChatGPT operates.
Gergana Vladova: For students, as well as adults, it's quite tempting to rely on ChatGPT for various everyday tasks like writing assignments, crafting texts, searching for information, or even putting together presentations. For one, this shows that there is too much trust in a technology that operates as a black box and doesn’t deserve this level of reliance. Secondly, depending too much on it can hinder the development of essential skills, like creative storytelling, independent research, and effective information filtering. To address this, we should create the right learning environments and ensure our teachers are well-prepared to guide students appropriately.
How sensible is it to prohibit generative language tools, or at the very least, limit their usage?
Gergana Vladova: I don't think it's possible to completely stop this development. It's a global phenomenon, and even if we restrict its use in the classroom, people can still use it at home for study or review. In my view, one of the essential skills for the future is being able to critically and independently navigate such technologies. That's why I support the idea of fostering these skills rather than imposing bans.
Sofia Schöbel: We are striving to become more digital, which means we need to grapple with artificial intelligence. ChatGPT illustrates how generative artificial intelligence functions. Instead of limiting ChatGPT, our focus in education should be on explaining its risks, highlighting its potential, and teaching safe and responsible usage from an early age.
Diana Knodel: These tools provide numerous educational advantages and can be a valuable asset in the learning journey. Rather than outright banning them, our approach should involve educating students about the possibilities and complexities of using such tools. Media literacy plays a pivotal role in achieving this. At the same time, it's essential to establish clear guidelines and ethical standards to prevent misuse or spreading of false information.
How can we address the uncertainties of parents and teachers in using these tools?
Sofia Schöbel: Just like students, parents and teachers should understand how to use ChatGPT. This means explaining the basics of artificial intelligence and showing them how ChatGPT works in action. Demonstrating how this tool can be a helpful addition to the teaching arsenal can be done through real-life examples and illustrations.
Diana Knodel: Communication and education are crucial here. Parents and teachers should receive thorough information about how language models like ChatGPT work and gain an understanding of their capabilities and limitations. At fobizz, we are dedicated to fostering an open dialogue and addressing questions to address any concerns. We provide training sessions for educators and host digital parent evenings on the topic of AI. We've noticed there’s huge interest from parents and teachers.
Gergana Vladova: For all of us, the primary condition for safe and self-determined engagement is building competencies. This means understanding how the tool works, recognizing both the opportunities and risks, and, importantly – because this is the role of parents and teachers – being able to convey this information to children effectively.
What risks do we need to consider in terms of data privacy?
Gergana Vladova: Data privacy remains a critical concern when dealing with all technologies. Personal data should be especially well-protected, and everyone – parents, teachers, and students – should be made aware of this. It's essential to have a secure environment in place for this purpose.
Sofia Schöbel: Users need to understand that ChatGPT learns from their inputs. There's already some awareness about this. Additionally, there's a risk that not all data entered is accurate, which could lead to incorrect or private information being incorporated into the tool.
Diana Knodel: ChatGPT from the company OpenAI is not permitted in German schools. Therefore, we need alternative solutions that anonymize queries and ensure that no personal data is required for use. That's precisely what we do with our fobizz solutions, always in close collaboration with data privacy experts.
How should we view ChatGPT in the context of an education system in crisis with a shortage of teachers?
Diana Knodel: An AI can certainly not replace teachers, but it can assist and support them in administrative and organizational tasks. In this regard, AI can play a role in addressing teacher shortages. However, it's essential to underline that human interaction and personal contact are irreplaceable in the learning process. Striking the right balance between technology and human support is vital to provide the best possible educational experience for students.
Sofia Schöbel: ChatGPT can complement teaching but won't replace teachers. It allows for a more personalized approach to learning and teaching, enabling teachers to tailor their instruction more individually to each student.
Gergana Vladova: Like in many other professions and applications, we can and should use the benefits of AI to provide support and enable more personalization, especially in overcrowded classrooms. However, it's crucial for teachers to acquire the necessary competencies to use technology effectively and address concerns. Even as better technologies emerge, we must remember that children view their teachers not only as a source of knowledge, and school is not solely a place for learning; it's also where they socialize and learn to navigate various situations, social challenges, and interactions with peers. For this, they need human role models and guides.
Thank you very much for the interview!
Diana Knodel is the founder and CEO of fobizz GmbH. A platform for advanced training courses and a provider of digital tools and AI for teachers and schools.
Sofia Schöbel is the Head of the Department of Business Informatics at the University of Osnabrück.
Gergana Vladova heads the research groups "Education for the Digital World" at the Weizenbaum Institute.
The interview was conducted by Leonie Dorn.
artificial&intelligent? is a series of interviews and articles on the latest applications of generative language models and image generators. Researchers at the Weizenbaum Institute discuss the societal impacts of these tools, add current studies and research findings to the debate, and contextualize widely discussed fears and expectations. In the spirit of Joseph Weizenbaum, the concept of "Artificial Intelligence," is also called into question, unraveling the supposed omnipotence and authority of these systems. The AI pioneer and critic, who developed one of the first chatbots, would have celebrated his 100th birthday this year.