🔎That's Some Deep Research

📚 Knowledge builders

• Deep Research → Open AI has enabled the Deep Research feature onto its free accounts. Limited to only 5 queries per month, everyone with an account will now be able to utilise the tool as it finds, analyses, and synthesises hundreds of online sources to create a comprehensive report like a research analyst. Powered by the upcoming OpenAI o3 model optimised for web browsing and data analysis, it uses reasoning to search, interpret, and analyse vast amounts of text, images, and PDFs on the internet. Here is an example of an output based on near and far transfer.

🤖 Industry updates

• From Teacher to Facilitator → This is not the first time that Alpha School in Brownsville, Texas, has caught my eye. It is pioneering a radical shift in education by replacing traditional classroom instruction with AI-driven learning. Their model compresses a full day’s academic content into just two hours using proprietary software developed by Trilogy Enterprises. The remainder of the day focuses on life skills and enrichment activities. I definitely have my scepticism, but the article is well worth a read.

• AI Education Executive Order → An executive order aimed at integrating AI into American education and workforce development. The initiative seeks to prepare students and educators for an AI-driven future, ensuring the United States maintains its global leadership in technology. The order focuses on four components:

  • White House Task Force on AI Education: A new interagency task force will coordinate national efforts to embed AI across the education system, led by top science, education, labor, and AI advisors.

  • Presidential AI Challenge: A nationwide competition will reward innovative uses of AI by students and teachers to inspire engagement and drive classroom adoption.

  • Educator Training and Resources: A scale up AI-focused teacher training and classroom integration, with NSF support for related research.

  • Workforce Development: The Department of Labor will expand AI apprenticeships and fund skills training through existing workforce programs to prepare Americans for AI-powered industries.

🪧Sponsored Deep Dive

Click here to see our principles around sponsorship.

A while ago, I set out seven principles that I think Intelligent Tutoring Systems need to follow to be effective tools. In short, these principles are:

1. Comprehensive curriculum crafted by experts

2. Scaffolded learning with gradual hints

3. Active recall and spaced repetition

4. Interleaving for enhanced conceptual understanding

5. Encouraging problem generation

6. Error correction and reflection

7. Cognitive load management

Third Space Learning have kindly allowed me to evaluate Skye, their commercially available AI Tutor, on these principles. This evaluation reflects the product at the time of writing. In the future, I have no doubt that the product will continue to evolve.

1. Comprehensive curriculum crafted by experts

   • Skye uses the curriculum and small steps set out White Rose as this was the structure that Third Space uses for their live tutoring sessions. White Rose, while far from perfect, is ubiquitous in primary schools, and it cannot be denied that the people behind it are primary maths experts.

2. Scaffolded learning with gradual hints

   • Skye use natural language to provide hints if a pupil gets an answer wrong or asks Skye what to do. This is certainly a big plus as it does not rob pupils of learning opportunities. However, these hints can sometimes be repetitive and not as fine-tuned to the errors that I have deliberatly made. In an early session, the session was around using adjusting but the scaffolded feedback I got to an incorrect answer related to using column addition. While far from ideal, it certainly is something that Skye could get better at.

3. Active recall and spaced repetition

   • The course that I have been placed on is a Year 6 Revision course, so there are plenty of mixed topics and opportunities for spaced repetition built into the course itself. A unique feature of Skye is that a teacher still needs to schedule a session. Without one, pupils cannot access tutoring. This means maximising spaced repetition and active recall is at the hands of the adult setting the sessions. As the system knows what pupils have been studying, I would like to see a feature that when pupils log-in to the system and no session has been booked, they can answer questions based on previously studied topics. This data should then be used so that Skye can adapt upcoming sessions as needed.

4. Interleaving for enhanced conceptual understanding

   • As the questions and lessons have been written by humans, aspects of interleaving have been used in questions. For example, the use of fractions when looking area and perimeter or the use of addition and subtraction in word problems. This is still an area that I would like to see Skye go further with, particularly during parts of independent practice where Skye could present questions that are related, but outside the targeted objectives.

5. Encouraging problem generation

   1. This is currently something that Skye is not able to do, but the team developing Skye is exploring ways to enable pupils to create their own problems in future updates. The range of possible responses and the specificity of the KS2 and KS4 curricula make providing appropriate guardrails challenging at present.

6. Error correction and reflection

   • Linked to the second principle, If a pupil gets stuck after three attempts, Skye explains the correct method in concise steps, ensuring pupils leave with clarity rather than confusion. At the end of each session, the pupils are encouraged to reflect on the session and use a Likert scale to provide some feedback on the session. Reflecting on the work that you have completed forces you to think hard, so it it good that the inclusion of this is available.

7. Cognitive Load Management

   • There is nothing worse than a busy slide, and Skye does a great job at managing attention by blurring out instructions and questions on the slide that the pupil should not attend to yet, and only reveals the interactive elements at the time when it wants an answer. This ensures that pupils are not tempted to skip ahead and not listen to the instruction. Furthermore, The lesson format – “I do, we do, you do,” followed by challenge questions – is consistent for each lesson, so pupils can focus on the substance, not on navigating a new structure every time. It would be good to see some variety of this, given what is known about the expertise-reversal effect and how too much guidance can hinder learning in some instances.

✨ Fresh prompts

1. Deep Research → It is not yet clear if there (at least to me) if prompting for Deep Research requires differently prompt frameworks to get the best answers for them, but here’s what I wrote for the Deep Research prompt around near and far transfer. I gave it some context, a detail request and some instructions. I find that it certainly gave me a strong starting point.
   
   Again, what is clear is that this prompt relied on me knowing something about the topic in question. It would be interesting to see how it fare when the user doe snot know much about the topic they are wanting to research.

Context:
Transfer in education refers to the ability of students to apply knowledge or skills learned in one context to another context. This can be either near transfer, where the context is similar, or far transfer, where the context is significantly different. Transfer is critical for lifelong learning and the application of academic knowledge in real-world scenarios. Research on transfer focuses on how teaching strategies, curriculum design, and student cognition influence this process. It has implications for curriculum development, instructional design, and assessment methods in education.

Request:
I would like you to explore the concept of transfer in education, discussing its various types, factors that influence it, and effective instructional strategies to promote transfer of learning. Please also analyze how cognitive load theory and variation theory might intersect with and impact transfer in the classroom.

Instructions:
	1.	Provide a detailed discussion of the different types of transfer (e.g., near transfer, far transfer).
	2.	Include an analysis of the cognitive processes involved in transfer, highlighting how cognitive load can either facilitate or hinder the transfer process.
	3.	Discuss how variation theory, with its emphasis on changing learning conditions and presenting content in different ways, can support transfer.
	4.	Use clear examples to illustrate the points where appropriate.
	5.	Structure the response into sections for clarity: Introduction, Types of Transfer, Influencing Factors, Instructional Strategies, and Theoretical Connections (including Cognitive Load and Variation Theory).