Reflecting on the use of AI in Learning and Development

I worked in marketing education for the guts of 10 years. During that time, for digital marketers the dream was to be able to “personalise” mass communications. To some extent you might argue that they could do this via a mail merge so that an email newsletter would start with “Dear Sean” but really the goal was to personalise the message based around an individual’s age, gender, interests, behaviour etc. For marketers, the goal is to influence an individual’s behaviour to get them to purchase whatever solution they are selling. Sometimes that can be a good thing if what’s being sold can add genuine value to somebody’s life. Or sometimes they’re just trying to sell more tat!

For educators, the potential for personalised education has always been intuitively understood but 1:1 tutoring is limited by economies of practice.

Understanding personalised learning

Personalised learning refers to an educational approach that tailors the learning experience to meet the individual needs, interests, and abilities of each learner. The goal is to enhance the impact of education by adapting teaching methods and modes to suit the unique characteristics of each learner.

The benefits of personalised learning include increased student engagement, improved learning outcomes, enhanced critical thinking skills, and generally greater self-confidence in learners because learning opportunities are always tailored to their individual ability. Personalised learning could bridge educational gaps because learners can receive targeted support in areas where they struggle. However, implementing personalised learning can be challenging, as it requires significant teacher involvement, access to appropriate technology and resources, and careful planning to ensure that all students receive the necessary support and attention.

Bloom’s 2 Sigma

The “Bloom’s 2 Sigma Problem” is a term coined by educational psychologist Benjamin Bloom in 1984. In a seminal research paper titled “The 2 Sigma Problem: The Search for Methods of Group Instruction as Effective as One-to-One Tutoring,” Bloom explored the disparity in learning outcomes between students who receive one-on-one tutoring and those who experience conventional group instruction.

Bloom’s research focused on the impact of different instructional methods on student achievement. He found that students who received personalised, one-on-one tutoring from a qualified and skilled tutor typically performed at least two standard deviations (2 sigma) better than students who were taught through traditional classroom methods. To put this into context, students in the tutoring group scored higher than approximately 98% of the students in the conventional instruction group.

The implications of the 2 sigma difference were profound, as it indicated the potential for improving learning outcomes if all students could benefit from personalised tutoring. However, providing individual tutoring to is impractical and resource-intensive proposition for most educational systems.

While Bloom’s work highlighted the potential for improving educational outcomes through more individualised instruction, finding practical and scalable solutions to achieve 2 sigma gains for all students remains a challenge for the field of education. However, his research continues to be influential in educational discussions and has prompted educators and researchers to explore new teaching methods and new technologies to try and narrow the achievement gap between personalised tutoring and conventional classroom instruction.

The impact of AI on Bloom’s 2 Sigma problem

The potential for technology to address Bloom’s 2 Sigma Problem has been a subject of ongoing exploration and debate among technologists and educators. Certainly, it would seem that we might be at an inflection point where AI has the potential to address some of the challenges posed by the 2 Sigma Problem by providing personalised and adaptive learning experiences. Here are some of the potential implications of the impact of AI on Bloom’s 2 Sigma Problem:

1. Personalised Learning: AI-powered educational platforms can analyse individual student performance data and learning preferences to create personalised learning pathways. By tailoring content and pacing to each student’s needs, AI can simulate aspects of personalised tutoring, potentially improving learning outcomes and reducing the achievement gap between one-on-one tutoring and conventional instruction.
2. Adaptive Assessment: AI could aeliver adaptive assessments that adjust the difficulty of questions based on the student’s responses. This approach ensures that students are appropriately challenged, motivated and guided toward areas that need improvement. As a result, students can receive more targeted and effective feedback to enhance their learning experience.
3. Teacher Support: AI can assist teachers by automating administrative tasks, providing insights into student progress, and suggesting appropriate interventions for struggling students. This allows teachers to focus more on individualised instruction and support.
4. Intelligent Tutoring Systems: AI can power intelligent tutoring systems that simulate the role of a human tutor by providing real-time feedback and guidance. These systems can analyse student interactions, identify misconceptions, and offer explanations or remedial content tailored to individual needs.
5. Data-Driven Decision Making: AI and data go hand in hand. As well as addressing individual learning challenges, AI can analyse large-scale educational data to identify trends. Educators and policymakers can leverage this data to make informed decisions about curriculum design, teaching methods, and resource allocation.
6. Inclusivity and Accessibility: AI-powered educational tools can be designed to accommodate diverse learning needs, including students with disabilities or those with different learning styles. This potential for inclusiveness can help address disparities in education.

I am reminded of Amara’s Law which states that we tend to overestimate the impact of a new technology in the short term and underestimate its impact in the long term. Time will tell of course how the use cases for AI in education will play out. In the meantime, I have found tools like ChatGPT to be very helpful in getting started on things such as generating initial course outlines, learning outcomes and even assessment methods. I have been able to create working drafts of these materials in minutes rather than hours which is amazing.

AI will increasingly be inflused in the tools that we already use.

In practice, I expect AI will increasingly be inflused in some of the tools that educators already use. I regularly use the AI assistant in Canva to speed up designing new slides. I suspect elearning authoring tools such as Storyline and Captivate will also infuse AI.

Ethical considerations

I’m an early adopter but I think it’s important to approach the integration of AI in education with caution. We need to beware of how tools process our data. We need to beware of algorithmic bias and of course, we need to maintain the humanity of quality education. Technology can be an enabler but nothing can replace the the skilled educator’s ability to build meaningful relationships with learners.

As AI technology continues to advance, it’s essential for educators, policymakers, and researchers to collaborate in designing and implementing AI-powered educational tools in a way that maximises their benefits while addressing potential challenges and ensuring equity and fairness.