AI Can Boost Education by Augmenting Teachers

At the Techonomy 23 conference on AI, Perpetual Baffour, research director of The Learning Agency Lab, presented her research on the integration of artificial intelligence in education. Her talk provided a balanced view on how AI could be applied in educational settings, emphasizing both its potential benefits and challenges.

Learning Engineering

Baffour highlighted the concept of learning engineering, an approach combining elements of computer science, learning sciences, and instructional design. This method aims to develop AI technologies that support effective teaching methods and learning processes. She described this as an initiative to ensure AI in classrooms aligns with educational best practices rather than being used for novelty.

The Feedback Prize Project

Baffour described how these principles were put into action with the Feedback Prize project, undertaken by her team. The project addresses the issue of writing proficiency among U.S. students, especially those from marginalized backgrounds. AI is used to provide personalized feedback on student writing, a strategy shown to improve writing skills. This project also aims to assist teachers with grading, offering AI as a tool to analyze student essays and provide feedback, thereby reducing the workload on educators.

Open-Source AI Tools for Education

An important aspect of the Feedback Prize project, as explained by Baffour, was the creation of an inclusive dataset of student essays, representing a wide range of demographics. This data was used to train AI models that were then tested and refined through competitions on Kaggle. The resulting AI algorithms, capable of evaluating student essays as well as a human, were made open-source, reflecting the idea that educational tools should be widely accessible.

Reckoning with Chatbots in Education

Baffour also touched upon the use of generative AI tools like ChatGPT in educational contexts. She noted the importance of customizing these tools to suit educational needs, highlighting potential biases and inaccuracies if used without modification. Her team worked on adapting ChatGPT to provide education-based advice, integrating it with a comprehensive database of educational research. This adaptation aimed to make AI more relevant and accurate for educational use.

Baffour ultimately expressed a cautiously optimistic view of AI in education. She emphasized the importance of syncing AI with educational research and practices and ensuring that AI tools are developed with equity and diversity in mind. Her perspective suggests that while AI has potential in educational settings, its application needs careful consideration and alignment with educational goals.

To learn more about Baffour and her work, see the article “How AI Can Tackle 5 Global Challenges” in the current issue of our sister publication Worth magazine.

(This summary was written with the help of generative artificial intelligence.)