E174 ‘Exploring AI in Education and Beyond’ with Professor Alan Smeaton

Today’s guest is Alan Smeaton, Professor of Computing at Dublin City University. and a member of the government’s AI Advisory Council. In this episode, Alan chats with host Mark Kelly about the evolution of Artificial Intelligence, its current applications in education and energy, and its future potential.

Alan has been at the forefront of AI in Ireland, reminiscent of a time when AI was neither fashionable nor widely accepted. His work in information retrieval and later in using AI techniques for audio, speech and video analysis has positioned him as a pioneering figure. He reflects on witnessing AI’s growth from rule-based systems and neural networks to the rise of generative AI, which captured global attention just two years ago.

Alan emphasises that despite the recent focus on generative AI, other AI forms remain relevant and active. He continues to integrate AI into his teaching, using it to enhance both the learning experience and teaching methodologies.

Alan then delves into role of AI in education, particularly in combating the traditional instructivist model. He advocates for a constructivist teaching approach where students learn independently, augmented by AI tools. His innovative use of a virtual teaching assistant – an LLM fine-tuned for his classes – demonstrates the potential of AI, allowing students anonymity and access to information 24/7, enhancing the learning experience.

This AI application not only streamlines teaching but empowers students to ask questions freely. Alan has observed the significant use of this tool, pointing out that students utilised it heavily, especially before exams, showing a reduced fear of asking questions they might consider trivial.

Addressing biases in AI data, Alan also shared his work on developing multiple LLMs trained to identify distinct biases, including gender, age and racial discrimination. By having these models work in collaboration, they offer responses free from singular bias, akin to a committee reviewing and refining answers. This approach highlights the importance of not relying solely on one AI model, advocating instead for diversity in AI usage.

Alan’s project on smart meters showcases another AI application, focusing on providing consumers with insights into their energy use. With 1.8 million smart meters deployed in Ireland, consumers can track their energy usage patterns. By uploading data to a website, users receive insights into the cost efficiency of various tariffs, potentially saving money through informed decisions.

Using AI to fill gaps in data, such as estimating annual usage from partial data, aligns with leveraging technology to offer personalised and actionable information.

Addressing the future of AI, Alan remains cautious about AI’s integration into business processes without proper grounding. He expresses skepticism about agentic AI handling complex tasks, as existing logical processing capabilities need further advancement.

Alan also discusses the AI industry’s landscape, expressing hope that it avoids concentration into a single dominant player, unlike the web search consolidation that left Google supreme. He encourages openness to multiple AI models, noting the dynamic potential of new players like Anthropic, with the Claude LLM, to enter the arena.

The episode concludes with a discussion on AI’s regulatory future and businesses’ need to engage in AI literacy. With over 100,000 products now embedding AI, the pressure to train employees on AI usage is evident. Companies must navigate regulation and ethical use, ensuring employees are equipped for a future increasingly intertwined with AI.

Professor Alan Smeaton’s contributions underscore the breadth of AI’s transformative power, from education to energy, while urging caution and diversity in AI’s ongoing evolution. This insightful conversation reinforces the imperative to embrace AI’s potential responsibly, ensuring broad access and benefits across all industries.