We're proud to announce that we recently won a grant from the Learner//Meets/Future AI-enabled Assessment challenge, organised by MIT Solve and funded by the Bill & Melinda Gates Foundation. We're one of just seven winners worldwide, which is both exciting and humbling.
But what does it actually mean?
Our AI approach: Modest & meaningful
We're not an AI company. In fact, we created Frankenstories in part because we couldn't get AI or NLP to work satisfactorily on Writelike.
We thought we might be a long shot for this grant precisely because our approach to AI is so modest. We're not building chatbots, story/essay generators, or formal assessment tools. There's no "Click here for AI magic!" button. ("Click the sparkles! Feel the goo!")
Instead, we're focused on using AI as a background support function to enhance what we already do well: getting students to write, a lot.
Both Writelike and Frankenstories encourage students to produce a significant volume of formative, exercise-style content. This is great for skill development, but it can create a challenge for teachers: how to effectively monitor and respond to all this writing?
This is where a delicate use of AI can help. We're exploring two main applications:
- Teacher sense-making: We want to give teachers a bird's eye view of student performance in both Writelike and Frankenstories. This includes assessing content quality, clarity, and (in Frankenstories) collaboration. The goal is to help teachers intuitively understand how their class is progressing so they can decide where to invest effort and how to focus their attention.
- Student feedback: In Writelike specifically, we're looking at providing immediate feedback to students on their responses to activities. This might be as simple as confirming they've understood the point of an exercise.
Why this matters
Our platforms are fundamentally about formative assessment and skill development. We're less interested in AI that scaffolds the output; we want to scaffold the learning.
And while foundation models such as Claude and ChatGPT can be great tutors for students who already know what they are doing, we're focused on the human part of learning, both peer-based and teacher-led.
We want AI to make it easier for teachers and students to navigate the complex process of learning to think and write in the complex, noisy, and multi-pressured environment of school.
What's next?
We're still in the experimental stages, and part of this grant's purpose is to help us figure out exactly how to implement these ideas effectively. Our first goal is to add a teacher-facing dashboard to Frankenstories that provides some insights at the student and class levels.
Stay tuned for updates! And as always, thank you for being part of the Writelike & Frankenstories community. Your feedback and support shape our work every day.