AI in education: Harnessing their potential and overcoming limitations

AI-powered tutoring has the potential to improve student performance and reduce academic disparities, which is particularly relevant in the aftermath of the COVID-19 pandemic. However, challenges remain, including limited teacher training, budget constraints, and addressing AI-powered cheating.

 Person, Reading, Advertisement, Poster, Text, Page, Clothing, Footwear, Shoe

The adoption of AI chatbots in education is gaining popularity, with a significant number of undergraduate students regularly using them for studying. Education specialists who harness AI are expected to have an advantage over generalist tech firms in the education industry. However, AI chatbots have limitations, such as providing inaccurate or unhelpful information, which education specialists can overcome by providing trusted and engaging content.

To ensure reliability, major textbook publishers like Pearson and McGraw Hill have chosen to train their own AI models instead of allowing AI chatbots to ingest their material. Pearson, for example, has designed AI tools that engage students by breaking down complex topics, testing understanding, and providing quick feedback. Meanwhile, Byju’s is personalising its AI tutoring tools by incorporating ‘forgetting curves,’ which refresh students’ memories at personalised intervals.

Tailoring AI chatbots to different age groups is crucial for an effective learning experience. AI chatbots must avoid confusing or simplifying the content, thus requiring education specialists to understand pedagogy. Established education suppliers have an advantage in integrating AI into familiar products and providing guidance to teachers on how to utilise AI effectively.

However, bringing AI to education is not without challenges. Many teachers still require training on digital learning tools, and budget constraints at educational institutions can hinder the adoption of new technologies. Overcoming scepticism towards AI and addressing AI-powered cheating are also major concerns that must be tackled.

The potential of AI-powered tutoring to improve student performance is supported by a study conducted in 1984 by educational psychologist Benjamin Bloom. The study found that one-to-one tutoring improves the average academic performance of students and reduces the variance between them. By making individual tutoring viable for many, AI has the potential to bridge the academic performance gap, especially among students from disadvantaged backgrounds.