Design

Design at Goodnotes: AI Math Assistance

November 17, 2023
Share

Chelsea Mangold-Takao is a Product Designer at Goodnotes who worked to bring AI Math Assistance to Goodnotes 6. Here’s how it’s made.

One of our goals at Goodnotes is to help students study and practice for high-stakes exams whether that be for the SATs, the DSE in Hong Kong, or the CSAT in South Korea. Studying for math presented an especially exciting problem for us as we know that solving math problems using handwriting is one of the most  common use cases in our app. We also knew from existing research that students who receive one-to-one tutoring performed 2 standard deviations better than students taught in a group setting like a classroom. This is called Bloom’s 2 Sigma problem, and we began to wonder how we could leverage our handwriting recognition technology combined with emerging AI tech to help students prepare for math exams through practice and assistance.

This is the goal we had in mind when we started working on our Interactive Exam Prep and AI Math Assistance features.

Introducing AI Math Assistance

From user research we knew that students still in the mastery phase of math were frustrated by how even the smallest errors could compound, leading them down the wrong path. Their final answer is wrong, but where in the process did the math go off track? For students still learning, this was difficult to pinpoint without help.

This is where AI Math Assistance can help.

Imagine having a trusted companion by your side, checking your handwritten math work in real-time, to provide timely feedback and prevent missteps from spiraling into confusion. In a nutshell, AI Math Assistance works similarly to spellcheck, but for math, by interacting with what you’ve written on the page. And just like other handwriting AI in Goodnotes, AI Math Assistance runs locally on your device so that all data stays private to you.

One of the challenges we faced was finding the balance between being helpful without diminishing the desired difficulty required for effective learning. We wanted to ensure that we were meeting students in their Zone of Proximal Development - the sweet spot in the space between what a learner can do unassisted and what a learner is unable to do, even with assistance.

To do this, we wanted to ensure that any assistance was unobtrusive, but enough to trigger self-reflection and self-correction without further assistance. For learners needing more guidance, we offer hints with explanations and step-by-step help. This way we are able to help students in different parts of their learning journey while fostering independence and helping to build critical problem-solving skills.

Balancing design and feasibility

In order to balance usability with feasibility, design and machine learning teams worked closely together. Research also informed us that students were revisiting mistakes in order to learn from them. Testing the prototype with students confirmed this when some learners chose to leave the mistake on the page rather than erase it and correct. They simply moved on to write the correction on the next line. We needed to ensure we could design the flow to take this into consideration.

Interactive Exam Prep

In addition to AI Math Assistance, the Interactive Exam feature comes equipped with a dashboard, that shows time spent on studying and your performance across the various topics. These statistics allow you to monitor your growth and celebrate your achievements.

AI Math Assistance is still an experimental feature, we’d love to hear your feedback. We currently offer practice content for the SAT, Hong Kong DSE and the South Korean CSAT, available to download for free under the Eduction section of the Marketplace Tab. Those of you in our free tier can test out AI Math Assistance with limited access to Interactive Exam Prep content, and our premium users have unlimited access to content and AI Math Assistance. If you’d like to see content for your local exam, let us know!

We cannot wait for you to give it a try and love it as much as we do. You can try out the new Goodnotes over in the Apple App Store.

If you want to reach out to our design and product team, don’t hesitate to tweet us @goodnotesapp with your requests so we can continue improving our beloved app.

Goodnotes uses cookies to enhance user experience and analyze traffic. Details of which cookies we use are available at our Cookie Policy. By continuing to browse the site, you accept cookies. You can withdraw your consent by adapting your preferences in the ‘preferences’ section.