Below are answers to some frequently asked questions about Unlocking Literacy. More information about the project, its phases and partners can be found here.

Still have questions? Please email us at literacy@theburnescenter.org

Who is running this project?

Unlocking Literacy is a project by the Museum of Science, Boston in collaboration with The Burnes Center for Social Change and Innovate Public Schools and with the support of Boston Public Schools’ Chief of Teaching and Learning. The project is supported by literacy experts at the Learning Agency, experts in artificial intelligence and machine learning from the Citizens Foundation, and undergraduate and graduate students in computer science, law and social sciences  from Northeastern University’s AI4Impact class and AI4Impact coop program.

What are the goals of this project?

By engaging diverse populations who are close to the problem, we want to better understand why improving literacy has proved to be such an intractable problem – both in Boston and nationally – and surface innovative and impactful solutions. In the short term, our goal is to create opportunities for underserved communities to meaningfully engage in public conversations around literacy. In the long term we hope to test, prototype, implement, and measure the effectiveness of promising solutions to address illiteracy in collaboration with partners. By incorporating artificial intelligence throughout the process, we hope to demonstrate the opportunities AI presents as a tool for equitable and effective problem solving.

What platforms and tools are you going to use?

We believe in using the right platform to serve the right purpose. We will use different tools, platforms, and problem-solving methods at each stage of the process.

To develop the content for our engagements, we use PolicySynth (Understanding the Problems) – PolicySynth is an open source project aimed at helping  governments and citizens make better decisions together by seamlessly integrating collective and artificial intelligence. Using PolicySynth’s AI-based tools, we will conduct web research into the root causes of low literacy and generate problem statements  for use in the All Our Ideas problem definition engagement.

For Stage 1: Identifying Problems, we use All Our Ideas, an  open-source survey tool for opinion gathering and ranking. The tool  presents respondents with a single question and then a random pairing of two answer choices. People select the statement they prefer (or “I can’t decide” as a third answer) or they may submit their own statement. At the end of this “voting” process, the tool outputs a rank-ordered list of the most popular statements. We will use All Our Ideas to ask the public to prioritize what they see as the important  root causes of persistent low literacy. 

For Stage 2 (Identifying Solutions) we will use Smarter Crowdsourcing – Developed by The GovLab, Smarter Crowdsourcing is a problem solving method which uses crowdsourcing to gather diverse ideas from global experts and rapidly develop those ideas into actionable proposals. Through a series of moderated, two-hour convenings on Zoom,  we will engage with literacy, policy, technology and other professionals to discuss how AI can help move the needle on literacy among those most chronically affected. Inside and outside of school.  The conversations are heavily facilitated to zero in on actionable ideas, rather than theoretical discussions. We will complement the collective intelligence of the group we convene with insights from artificial intelligence.

For translations, we use a combination of WeGlot, ChatGPT, and human translation services.

How are you using artificial intelligence?

Generative AI makes it easier to generate and to organize information. As a result, it can help us and every participant summarize and organize what is being said. The AI-based tools mentioned above will be used to generate content to support the engagements and to aid in the facilitation of the engagements. In addition, we will use tools like ChatGPT throughout the project to help with summarizing, processing, and drawing insights from the outputs of the engagements. We will also use open source machine translation tools like Seamless M4T to translate materials into languages other than English to ensure that diverse language communities are able to contribute their insights.

What guardrails are in place to make sure the AI produces good results?

We employ a “human-in-the-loop” model of AI deployment. In other words, none of the generative AI tools we use are fully automated processes. While AI is used as a starting point to generate content more efficiently and from a greater variety of sources, real people review all outputs for accuracy and comprehensiveness. The content we produce with the help of generative AI will be reviewed by education experts from the Burnes Center as well as literacy and learning experts from the Learning Agency. Further, where AI is used to aid with translation, fluent speakers of each language will review the written content for accuracy and clarity of understanding. 

How will you ensure that folks with low literacy are able to make their voices heard?

The online, text-based engagements (which require moderate to high levels of literacy to participate) are complemented by a series of in-person, spoken engagements where we will prioritize engaging those with low literacy. These include focus groups, workshops, hackathons and other types of face-to-face meetings.