How Can AI Support Teachers and Personalize Learning?

TL;DR

AI should strengthen classroom teaching by personalizing learning, answering questions, and showing teachers where students need help, while leaving educators in control. Human teachers remain essential as mentors, role models, and sources of social connection, so responsible educational AI should work within existing systems instead of replacing classrooms with screen-based instruction.

Transcript

[MUSIC PLAYING] HANNAH FRY: Welcome back to Google DeepMind, the Podcast. I'm your host, Professor Hannah Fry. Now, there are few places in the past couple of years that have felt the transformative influence of AI as keenly as the education sector. Teachers around the world are having to abruptly rethink how they engage with and assess students, a... Read More

Key Insights

• AI in education is best treated as an addition to existing classrooms, not a replacement for traditional schooling. Educational systems are longstanding, established structures that combine knowledge transfer with social development, peer learning, critical thinking, information evaluation, and preparation for participation in society.
• Past educational technologies have not fulfilled predictions that they would overturn conventional education. Intelligent tutoring systems have existed for more than 50 years, while massive open online courses often serve people who already completed traditional education, including learners seeking an additional master's degree.
• Human teachers are essential because teacher-student relationships contain commitment, social expectations, inspiration, and personal connection. A learner can instantly close an AI tutor, but interacting with a human tutor creates a different form of engagement that technology cannot fully reproduce.
• Google DeepMind's ambition is to make Gemini the best large language model for education while supporting multiple learning experiences. Its researchers want educational communities with deeper experience in the field to help shape how AI tutors work, rather than having one organization dictate their design.
• An AI tutor on YouTube learning videos lets viewers virtually raise a hand when they do not understand something. The tutor appears within the learning experience and answers questions, giving students immediate assistance at the moment when confusion occurs.
• The Gemini learning coach is designed to guide open-ended learning about subjects selected by the user. It can propose a plan, identify what the learner appears to know or not know, and lead the learner through relevant material instead of merely supplying an isolated answer.
• AI-first schooling is not the future the researchers advocate because students should not spend the school day simply looking at screens. Their preferred model keeps teachers and peer interaction central while using AI to help teachers and learners improve the classroom experience.
• Personalized AI support could let each learner progress at an appropriate pace and pursue individual interests. At the same time, teachers could see where students are, steer the tutor, preserve their own teaching styles, and devote more attention to learners who particularly need personal help.

Questions & Answers

Q: How should AI be used in classrooms without replacing teachers?

AI should operate as a supporting tool inside the existing educational system. It can personalize instruction, answer questions, help learners move at their own pace, and provide teachers with information about student progress. Teachers should retain control over lessons, preserve their teaching styles, and continue serving as mentors, role models, and sources of human connection.

Q: Why are human teachers still important when AI tutors are available?

Human teachers contribute far more than factual instruction. They inspire interest in subjects, teach critical thinking and information evaluation, model how to participate in society, and create meaningful relationships with students. Human interaction also carries social expectations and commitment, whereas a student can simply close an AI tutor whenever the learning experience becomes difficult or unappealing.

Q: Why have earlier educational technologies not replaced traditional education?

Earlier technologies entered educational systems without eliminating the broader functions of schools. Intelligent tutoring systems have existed for more than 50 years, yet their full promise has not materialized. Massive open online courses also failed to replace conventional education and often attracted people who had already completed traditional schooling, including those pursuing another master's degree.

Q: What does the AI tutor on YouTube learning videos do?

The YouTube feature supports viewers while they watch learning videos. When a learner does not understand part of the material, they can virtually raise their hand and open an AI tutor. The learner can then ask questions related to the lesson, receiving assistance within the learning experience instead of having to abandon the video or search elsewhere.

Q: How does the Gemini learning coach guide a learner?

The Gemini learning coach accepts a subject or learning question, such as a request to study photosynthesis or the American Civil War. It creates a plan, tries to determine what the learner already understands and where knowledge is missing, and then guides the learner through the material. Its purpose is to structure an ongoing learning experience.

Q: What would a responsible AI-supported classroom look like?

A responsible classroom would keep teachers, students, and peer interaction at its center. AI would help individual learners progress at suitable speeds and explore their interests, while teachers would see where everyone stands and steer the tutoring system. Educators would maintain control, bring their personalities to lessons, and intervene when students need direct human support.

Q: Why does Google DeepMind want educators to shape AI tutors?

Google DeepMind does not want one organization to dictate what an educational AI tutor should be. Its researchers hope to listen to people who have worked in education for much longer and help them use the technology effectively. Community involvement can inform both improvements to the underlying models and decisions about how those models are applied in learning experiences.

Q: What risks come with AI-first schooling?

AI-first schooling can reduce education to students sitting in front of screens while weakening the teacher relationships and peer interactions that make classrooms valuable. The researchers do not view replacing human teachers as a desirable goal. They favor AI that reduces difficult situations, supports overworked educators, and creates more opportunities for memorable, inspiring exchanges between teachers and students.

Summary & Key Takeaways

• Educational technology should augment established classrooms rather than attempt to dismantle them. Earlier intelligent tutoring systems and massive open online courses did not replace traditional education, partly because schools provide more than information. They also cultivate peer relationships, critical thinking, information evaluation, citizenship, mentorship, and meaningful connections between teachers and students.

• Google DeepMind aims to improve Gemini as an educational model that can support different learning experiences. Examples include an AI tutor attached to YouTube learning videos and the Gemini learning coach. These systems can answer questions, assess what learners know, propose learning plans, and guide students through unfamiliar material at an appropriate pace.

• The preferred future retains teachers as mentors, role models, and active directors of learning. AI could personalize instruction, reflect individual interests, and give teachers visibility into each learner's progress. Teachers could steer the system while preserving their personalities and teaching styles, freeing attention for students who most need personal support.