As part of our generative AI project, we’re building a handful of prototypes to demonstrate how people in the early years sector could make use of the benefits of large language models such as ChatGPT.

We're also exploring how to make our prototypes easy to access through apps that early years practitioners already use, such as WhatsApp.

We're publishing our work as we go, so that other developers can use our processes to build their own chatbots for social good.

You can find our technical blogs on our Medium publication, Discovery at Nesta.

They include:

• Using Streamlit and GPT-4 to prototype a web app for suggesting personalised early-years activities
• Combining WhatsApp with large language models: prototyping with Twilio and Flask
• How to use GPT-4 and OpenAI’s functions for text classification
• How to build an early-years activity recommender with retrieval-augmented generation, GPT-4 and Pinecone
• How to evaluate large language model chatbots: experimenting with Streamlit and Prodigy

If you'd like to know more about our work, get in touch via email.