ChatGPT in education
Professor Rose Luckin (IOE, UCL's Faculty of Education and Society) explains how the chatbot is transforming learning
People | Portico Magazine | Autumn 2023
I have had the privilege of witnessing the remarkable and occasionally controversial evolution of ChatGPT since its launch in Autumn 2022.
Having spent over 30 years studying AI and its application to education, I am delighted that finally AI is being taken seriously by schools, universities and colleges. But the way that the story of this technology and its application in education and the workplace is unfolding demonstrates that we need to be thoughtful and strategic in our response.
ChatGPT is an advanced AI language model that generates text of human-like quality. It was developed by a company called OpenAI and builds upon the success of its predecessors, leveraging deep learning techniques and vast datasets to generate its sophisticated responses. It is a type of AI known as Generative AI, which includes other tools such as Claude (another text generation tool) and Stable Diffusion (an image generation tool).
ChatGPT operates through a neural network architecture known as the transformer model. This means that it excels at processing and generating text, making it capable of engaging in conversations, answering questions, and even offering creative solutions. Its ability to adapt to various tasks and contexts has made it an incredibly versatile tool.
Since its release last year, it has been on a considerable journey of development and will soon be able to generate custom images as well as text. ChatGPT is already being applied in education, and is also a component in a wide range of new products and services being planned and released onto the market each week.
"One of the most remarkable aspects of the development we've seen is its speed."
One of the most remarkable aspects of the development we've seen is its speed. With each iteration, ChatGPT becomes more proficient at natural language understanding and generation. The research and engineering teams behind the platform have worked to fine-tune its capabilities, making it increasingly valuable in a wide range of applications.
However, ChatGPT has no understanding of the way that its outputs apply to the real world and this deficiency brings limitations to its usefulness. It is also prone to making things up, and therefore its answers are not always accurate. Not surprisingly, it has been the cause of considerable criticism and anxiety.
How is ChatGPT being applied in an educational setting? How does it complement other learning technologies and traditional teaching methods?
ChatGPT has quickly found an audience in education. It is therefore important to contextualise its potential to support teaching and learning within the broader range of AI tools about which there has been decades of learning.
Research has demonstrated that AI can be used in education and training and I know from my own experience that these sophisticated tools can complement other learning technologies, such as online classrooms and augmented reality applications.
"Personalised learning into the hands of millions more educators and learners."
AI, like the Natural Language Processing that enables speech to text and the Machine Learning algorithms that power existing adaptive learning tools, can enable technology to adapt to individual students' needs and learning pace, providing customised explanations and practice problems. This personalised approach can enhance each learner’s experience.
ChatGPT adds a new tool to this existing AI repertoire and the fact that it is available at scale brings the potential for personalised learning into the hands of millions more educators and learners.
An AI, like ChatGPT, can also provide 24/7 support for students, offering them the opportunity to seek help outside regular teaching hours. Its use can promote independent learning and problem-solving skills.
I have also seen AI, including ChatGPT, being used effectively to complement traditional teaching methods by offering personalised and on-demand assistance to educators. Teachers can utilise ChatGPT to gather and create resources, plan lesson, and even receive assistance with administrative tasks, allowing them to focus more on teaching itself.
What are the benefits of incorporating ChatGPT into our work?
As with many AI tools, the integration of ChatGPT into education offers a range of potential advantages, with students benefiting from access to a wealth of additional educational resources, and from the engaging and diverse lesson plans that teachers are more easily able to create. Plus, of course, the more personalised feedback from the technology combined with the increased ability for teachers to address their specific learning needs effectively.
More specifically, I have found ChatGPT to be useful for language translation, grammar correction, and pronunciation guidance, and I know others too are finding it a helpful companion for foreign language learners. There are also accessibility benefits to ChatGPT and some other AI applications, which can provide features such as text-to-speech and speech-to-text, ensuring that students with diverse learning needs can participate fully.
I am also convinced that the right combination of AI technology and expert human teaching can enhance and simplify the way that complex concepts can be explained, fostering better understanding and retention.
What are the risks associated with ChatGPT in education? How do we guard against them?
The potential and the enthusiasm must of course be tempered by caution. While ChatGPT holds immense promise, it also comes with considerable risks, particularly when it comes to its use in an educational context.
I have already highlighted in the introduction that ChatGPT gets things wrong and its answers do not represent any understanding of the real world.
"Continuous monitoring and bias mitigation strategies are essential."
In addition, we also need to be mindful of privacy worries, especially when handling student data. Stringent data protection measures and policies must be in place to safeguard sensitive information. There are also reasons to be concerned about bias and misinformation, because AI models like ChatGPT can inadvertently perpetuate biases present in the data they were trained on. This means that continuous monitoring and bias mitigation strategies are essential to ensure fair and accurate information.
Personally, one of the areas about which I have particular concern is overreliance. There's a risk that students may become overly dependent on ChatGPT for answers, potentially hindering critical thinking and problem-solving skills. To guard against this, educators must encourage independent thinking and emphasise the tool's role as a supplement rather than a replacement.
Where do we go from here? What should we aspire to do with technologies like ChatGPT?
The future with ChatGPT and similar AI technologies in education is full of promise, but to harness their potential fully, I believe we need to set ourselves high standards. We must aspire to:
- Continuous Improvement: AI models should be regularly updated and improved to offer increasingly accurate, unbiased, and helpful responses;
- Ethical AI Education: Incorporating ethical AI education into curricula will empower students to use these technologies responsibly and understand their societal impacts;
- Research Collaboration: Collaboration between AI researchers and educators is essential to tailor these technologies to specific educational needs effectively;
- Equity: Efforts should be made to ensure that AI-powered educational tools are accessible to all, regardless of socio-economic background.
In conclusion...
The integration of ChatGPT into education represents a transformative step forward. Its ability to provide personalised support, enhance accessibility, and streamline administrative tasks has the potential to revolutionise the education landscape. However, revolutions are invariably a bumpy ride!
"Immense promise and profound challenges."
The future of generative AI holds immense promise and profound challenges. It is a journey into uncharted territory, where we must tread cautiously, guided by a commitment to transparency, ethics, and responsible AI education.
As we look forward, we are tasked with the formidable mission of striking the delicate balance between harnessing the power of AI to enhance our lives and safeguarding the essence of what makes us uniquely human – our cognitive capacities, our adaptability, and our capacity for metacognition and self understanding.
The path forward for human learning and evolution lies in the fusion of AI Literacy, Learning Mastery, and Knowledge Mastery, creating a foundation upon which we can navigate the complex and evolving relationship between humanity and artificial intelligence, and always make sure we stay one step ahead of the ‘robots’.
Rose Luckin is Professor of Learner Centred Design at the UCL Knowledge Lab within IOE, UCL's Faculty of Education and Society.
Portico magazine features stories for and from the UCL community. If you have a story to tell or feedback to share, contact advancement@ucl.ac.uk
Follow UCL
Built with Shorthand