SCAI Question 8

Transforming Education

How can we use AI to enhance the effectiveness, efficiency, and accessibility of education across societies around the world?

Context & Assumptions

Improvement of human capital through education is foundational across society and economy. There is great potential for AI to enhance the efficiency, effectiveness and accessibility across the entire educational ecosystem. This spans students, parents, teacher training, content creation, social interaction, curriculum design and delivery, evaluation and certification of results.

An AI-enhanced education ecosystem can cater to a range of needs across countries and contexts. In resource-constrained societies, this could mean the difference between education or no education at all. In other cases, knowledge and skills education could be enhanced to free up resources to address the more “human” aspects of education, such as critical thinking, creativity, empathy, social development etc, as well as focusing on student-teacher interactions. New models of human-AI collaboration could be explored, such as in using AI tools to assist teachers in characterising and evaluating the students and their progress, so as to challenge them appropriately and foster learning and personal growth. Across industry, appropriate interventions could improve upskilling and the adoption of new tools and capabilities, enhancing productivity and introducing new value.

This approach relies on several assumptions. At the core, good access to infrastructure and connectivity are key; every student needs access to a mobile phone and a TV, possibly a keyboard. On the AI side, we assume the current state-of-the-art for AI, i.e. there are few unsolved scientific AI challenges that prevent us from making progress in this area and we can see payoff and test product-market-fit from the start.

Question

How can we design and implement an AI-enhanced, open education ecosystem, with a sustainable mechanism for participation, with the ambition of making education more efficient, effective, and accessible for local communities and global society?

Doing so will also enable us to maximise human capital across communities around the world, starting with maximising every student’s potential.

Indicators of Progress

Given the complexity of education and the contextual needs of communities and corporations, a systems approach is necessary. This recognises that opportunities and incentives differ across life phases–elementary school, tertiary education, workforce reskilling, and more. Moreover, we need to address all elements of the value chain: students, teachers, curriculum optimization and delivery, content creation, evaluation, certification, accessibility and social interactions. The following are non-exhaustive illustrations:

Students can benefit from having an always-available tutor, and benefit from immediate feedback and personalised, adaptive content.
Parents care about supporting and monitoring their childrens’ progress. AI can help them support them, such as in pointing out areas of improvement and suggest actions.
Teachers will benefit from spending less time on routine tasks such as grading, and can devote more time to foster creativity among students.
Content capture and creation is time consuming, thus a major hurdle for educators to share courses. AI can be used to automatically transcribe and generate material.
Curriculum design and optimization currently relies on the experience of lecturers, drawn from hundreds of students. A data-driven approach based on 100x more student experiences can tailor curriculum for better learning, customised to individual students.
Evaluation involves certification (of achieved degree), challenge (for progress) and control (of education objectives). Game mechanisms combined with empirical data can motivate students through suitably challenging problems and tests.
Accessibility tools are typically costly (subtitles, speed, visual aids); these can be added through AI tools at minimal cost.
Student interaction can be enhanced by social recommendations (student / tutor pairings, peer groups) and content moderation.

There is significant opportunity for productivity gains. These dividends can be used to increase the number of students trained, and/or the quality of their education, for a more cost-effective education system.

To achieve this, we must overcome several challenges, including:

Overcoming accuracy issues such as hallucinations.
Integration with existing infrastructure such as Google classroom.
Interoperable, universal software interfaces that allow for whole system optimisation (future-proofing vs. ease of use vs. adoption), and that allows for meaningful auditability by humans; text-based interfaces are likely most suitable.
Educator capabilities in using AI in teaching.
Contextually-aligned curriculum and delivery depending on cultural norms and needs.
Having strong enough economic incentives to facilitate diverse offerings and interoperability in an AI-enhanced education ecosystem.
Regulatory and commercial acceptance of certificates obtained by AI-enhanced education (in particular for self-study and non-traditional pathways).
Social acceptance of the notion of AI-enhanced education itself, by stakeholders such as educators, parents, students, companies, regulators.