Guidance

This website is a place for ideas, guidance, shared knowledge and open-source code and tools to help the education community learn how to sensibly integrate their work helping children learn.

On this page we’ll share guidance that we’ve developed for our community of educators and developers. From our introduction to AI – a detailed guide written in simple language – to more technical guides about Retrieval Augmented Generation (RAG).

Introduction to AI

Artificial Intelligence can seem overwhelming, given the pace of the advancement and the complexity of the methods used. We at AI-for-Education.org believe that while how AI works is complex, the core principles of AI should be accessible to all – especially for those who are using these tools to help children learn.

To help, we’ve developed this introduction to AI – which walks you through the main definitions, main techniques, and the history of how we got where we are today. It’s as non-technical as possible in this field – but an important read for all of you who will build or use AI tools in the future.

AI Prompt Writing Guide for Educators

Prompts are the questions we write into these models, usually through the web interface or apps. The clearer, and more detailed you are with your prompts, the better the outputs – if you think about how you’d ask a teaching assistant for help – being clear with your goals, setting things out step by step – then you’ll prefer the results.

This guide presents some simple examples of what LLMs can do and strategies you can use to get what you need when using them, particularly through improving your prompts. It also shares a little on what they are not good at – and some things to be cautious of.

Retrieval Augmented Generation

“How can we use our content with these AI models? We need them to speak to us, not just western ideas”

This is the number one concern we hear from Educators and Ministries across low-resource countries. They are impressed by AI models, but then hear about ‘learning styles’ or see math when it should be maths, and instantly wonder how it can be made for them.

Fortunately, the last year has seen a real series of innovations in this space – with many options now available to developers/ tool builders in LMIC’s. One of the highest potential is Retrieval Augmented Generation – RAG. Here, you combine the AI tools with databases of content so you use your existing knowledge.

We’ve been exploring this with the team at RTI International and Science of Teaching experts – looking at how we can build a model which uses this high-quality inputs.

As part of it, we wrote down why we doing this, and what we are doing, to show others how they can use the same techniques to improve their apps. To find out more about how RAG works, we put together this handy overview.

AI Benchmarks for Education

As AI use for education proliferates, our priority is to ensure that AI tools are high quality. This means thinking about evidence – and in AI, about ‘benchmarks’. While there are many benchmarks for testing if AI systems can pass exams, there’s nothing out there to test if AI systems can help teachers or help kids to pass exams.

This month we brought together people thinking about this, and released this note sharing what benchmarking means in our context, and how we will shape our work on developing education benchmarks over the coming months.

We are realising that AI in education is about upgrading people and processes – our traditional ways of working are being updated – and as a result we need new ways of thinking about evidence. Over time, this will evolve into understanding what the ‘smart buys’ are for AI in education.

How AI can help with system strengthening in education

Education is a system – so we need to also think about the non-teaching workforce, and the tasks they do – when we talk about AI-for-Education.

To do this, we worked to map out “what do people do all day” across the system – we came up with a long list of tasks which we then tried to summarise for common patterns and ideas for potential upgrading with AI. And we found lots – and potential challenges, and things to do before we can use AI sensibly and safely.

Gen AI – What does it mean for education?

We were recently invited to talk with the Ministry of Basic and Senior Secondary Education in Sierra Leone about what Generative AI will mean for education.

We started with a basic introduction – reminding everyone that generative AI models are just really good at predicting words; but that if the words they’ve read are not from your context, then you might know better than they do.

The overarching feeling was one of optimism, and a need to keep an open-mind – but a real need to work to make sure these models build on the content and knowledge developed by the Ministry, and that children’s and teachers privacy is protected.

This guide presents some simple examples of what LLMs can do and strategies you can use to get what you need when using them, particularly through improving your prompts. It also shares a little on what they are not good at – and some things to be cautious of.

Here are the slides we shared on the day.

What does AI mean for Education Ministries?

We went to Education World Forum 2024 where we spoke to many education ministers about AI – where it might be used for education, innovations, what ministries need to know and the questions ministries should be asking.

We thought this handout might be useful to others.

How AI Can Help With Education

AI has the potential to transform approaches in education, both inside and outside the classroom. Here, we explore some suggestions of where and how AI technology may be used throughout the education system.

We have established 19 areas of the education system where we envision the use of AI and for each of these areas we have described potential appliations. By outlining these applications we hope to frame dialogue within the community and to establish priorities for investment and development.

Learning By Doing

We are providing small grants to support the development of AI products & components in LMICs. We know that innovation investment is high-risk. Our aim is that our community can benefit from the lessons learned in these pilots – what works and what doesn’t.

Learn more about our pilot projects here. We will be following each project and reporting on key learnings.

About our GitHub

The AI-for-Education Github organisation will be a hub for developers, hosting open-source code repositories and discussion forums. As well as being the home for all code associated with AI-for-Education’s grantees, the organisation also welcomes contributions from any interested parties working on foundational literacy and numeracy applications in low and middle-income countries, subject to an approval process.

To kick-start this process, the organisation currently hosts two LLM toolsets used by Fab Data‘s Teacher.AI products:

  1. fabdata-llm is a high-level python interface to multiple different LLM providers and providing a simple chat management system;
  2. fabdata-parsedoc is a python library for LLM-powered document I/O, cleaning, and content processing.

We look forward to your contributions!

Sign Up

Join our mailing list to keep up to date with news and events.

Community          Knowledge          FAQ
          Privacy Policy

AI-for-Education.org was set up by Fab Inc. in partnership with Team4Tech. We are grateful to the Bill & Melinda Gates Foundation and the Jacobs Foundation for their support.

Powered by FabData.IO