 This video presents a basic way to use “retrieval augmented generation” (RAG) with your own Google Drive as a knowledge source. Plug-and-play, it uses Google Colab, which can be accessed for free, and Mistral AI open source models, with API calls that should only cost a fraction of a cent per run.
0 Comments
 A glimpse into the future of the future! AI integration in a school context is still too focused on immediate applications, and thus on the consumption of task-specific tools (mostly designed by non-educators to automate traditional practices). The greatest potential of these technologies, however, comes from the fact that they make it possible for educators to create and share their own solutions. Here is what this could / will look like.
 Intriguing study by MIT Professor Markus Buehler, who presents original ways in which AI technologies can help accelerate scientific discovery through knowledge graphs, creative thinking, unterdisciplinary approaches, including artistic inspiration, and philosophical insights.
 I was fortunate enough to be given early access to StretchAI, a “digital coach… trained to answer questions about effective teaching and learning practices…”.
 As the name “GPT” indicates, generative models have been “pre-trained” on large data sets, allowing them to learn general patterns in human language and pick up knowledge and skills in the process. Only accessible to experts a year ago only, post-training is now easy to implement through plug-and-play platforms, promising to specialize LLMs in complex and specific tasks such as essay feedback.
 Google’s Gemini 1.5 comes with multimodal capabilities and a 1M-token context window. This makes it possible to chat with very long videos, such as class recordings, and thus to use the model as an observer, data collector, and instructional coach.
 Concerns around the reliability and equity of generative AI are particularly salient in a school context. This short video shows how bots can be made to “think” critically through the collaborative work of a team of agents.
 Task-specific tools are nice to catch people’s interest and get them started with AI, but they are only an entry point. The true potential of these technologies, including in education, will come from the development of AI agents teams equipped with advanced functionalities and responsible for complex, automated workflows. Here are two simple, early examples.
 When one of our learning support teachers asked me if there were any AI tools that would take notes for students, I could think of several solutions…but decided to create one myself and use this as an opportunity to test whether open-source resources (which help solve issues of privacy and cost) would be performant enough to be fit for purpose.
 AI enhancements are often geared towards older students, but Microsoft's new Reading Coach promises to engage younger learners and help increase reading fluency by allowing them to co-create stories, adjusting reading levels, providing scaffoldings, personalizing practice, celebrating their achievements - and providing useful data to educators. There is a number of ways Reading Coach could be further improved, but it is free and safe to try.
|