Considerations for Artificial Intelligence (AI) and Human-Centered Design

Written by Jody Britten, Ph.D. | Team4Tech Foundation | Basic Education Coalition member

Jody is a member of BEC’s EdTech working group. Stood up in 2020, this working group has held regular technical exchanges with BEC’s members to share best practices, challenges, and lessons learned from the EdTech space. The group has collaborated on conference presentations and panels, organized and shared at Fail Fests and webinars, and carried out a BEC-wide EdTech survey “Governments’ Organizational Responses to COVID-19”.

Generative Artificial Intelligence (AI) has significantly impacted global education in the past year. From advanced algorithms and personalized learning materials to data analysis and automation of administrative tasks; generative AI has brought positive changes to education. Even with all the advances and opportunities, there is still work to do to ensure that generative AI is developed and used ethically and transparently.

Team4Tech is a nonprofit impact accelerator that leverages technology to drive positive change in education. By combining the power of technology, education, and collaboration, we use human-centered design to ensure technology solutions are developed with the aspirational needs of our network of education-focused nonprofit organizations at the forefront. Through our collaborative efforts, we have successfully implemented training and created resources that help global partners harness the potential of generative AI to enhance learning experiences and support educators.

Over the past ten years, we have witnessed first hand how human-centered design can play a crucial role in supporting the inclusive design and development of educational technologies. As a multidisciplinary approach that places user needs, behaviors, and experiences at the epicenter of the design and development process, human-centered design is rooted in a deep researcher base that goes beyond mere usability to foster meaningful interactions and sustainable engagement.

With members of our community of practice representing more than 85 countries and 500 unique non-government organizations (NGOs), Team4Tech has engaged our community in meaningful conversations and training opportunities around generative AI. Through that process over the past six months, we have identified five salient considerations for developers and users alike which have the potential to positively impact the use, design, and development of these tools as we move forward.

• Consideration 1: Ethical Implications of Autonomy

• Incorporate ethical considerations from the outset, bringing accessibility and equity to the table during initial design and pro-typing. 

• Utilize frameworks like Ethical-by-Design or Value-Sensitive Design to support design teams that lack global experiences or perspectives.

• Address questions about agency, consent, and data privacy and be transparent with where, how, and when data is shared, stored, and used for AI training.

• Consideration 2: User-Centric Personalization

• Balance personalized experiences with user cognitive load, cultural context, and privacy expectations.

• Ensure pathways to delete data and validate personalization techniques.

• Prioritize equity in personalization practices, being diligent and transparent about how users can customize, adapt, and train generative AI tools to better meet their needs.

• Consideration 3: Interdisciplinary Collaboration

• Adopt an interdisciplinary approach to integrate generative AI with human-centered design.

• Include domain experts in machine learning, UX design, ethics, and behavioral psychology.

• Foster open community conversations about AI by involving diverse stakeholders and generate opportunities for feedback loops with end-users.

• Consideration 4: Algorithmic Transparency

• Strive for transparency in decision-making processes without compromising intellectual property.

• Address questions about algorithm training and the inclusion of diverse global audiences.

• Test algorithms with specific audiences in mind and share the test-user demographics to demonstrate attention and awareness of bias.

• Consideration 5: Usability and Accessibility

• Conduct rigorous usability testing to ensure broad accessibility and transparency in how population differences have been accounted for in AI training.

• Prioritize accessibility features and involve end-users in collecting continual feedback.

• Routinely address the data and power consumption that is needed to use and build generative AI tools and applications.

As we continue to navigate the intersection of generative AI and human-centered design in education, it is crucial that we foster an ongoing dialogue and collaboration among users and developers. By working together, we can ensure that generative AI tools are designed and used equitably, promoting inclusive learning experiences for all.

To contribute to this dialogue, users and developers can engage in open conversations and advocate for ethical guidelines. Participate in online forums, communities, and conferences dedicated to discussing generative AI and human-centered design in education. Share insights, ask questions, and learn from others to enrich the collective understanding of equitable design practices. And encourage the development and adoption of ethical guidelines and standards for generative AI in education. Advocate for transparency, fairness, and inclusion in AI algorithms and decision-making processes. By promoting ethical considerations, we can shape the future of generative AI to prioritize the well-being and empowerment of learners and educators.

Team4Tech is continually exploring the possibilities of generative AI while ensuring that it aligns with the principles of human-centered design and equity. By embracing these important considerations and taking action, we are doing what we can to pave the way for a future where technology truly enhances education for everyone.

You can see cataloged resources focused on AI in global education that are continually curated by Team4Tech’s Director of Learning Technologies here.

BEC Celebrates the Global Day of Parents

In honor of the Global Day of Parents (June 1, 2023), let’s take a look at research by BEC member Food for the Hungry (FH) that analyzes communities strengthened through education across 10 countries. Below, FH’s Lisa Easterbrooks, Director of Education Programs, outlines the most recent findings on their community-based “care group” interventions.

The United Nations reported that 1.6 billion learners in more than 190 countries were affected by the closure of educational institutions at the peak of the COVID-19 pandemic. Between February 2020 and February 2022, educational systems were fully closed for in-person schooling for an average 141 days.

As a result, the World Bank estimates the percentage of children who cannot read a simple story by age 10 has increased to 70% since onset of the pandemic. All progress that was achieved toward mitigating learning poverty in low- and middle-income countries since 2000 has been lost.

Millions of children and youth now face the challenge of catching up on lost learning. Food for the Hungry (FH) is working with caregivers, teachers, and communities to improve child learning and development through social behavior change interventions.

FH recognizes that school-based approaches alone are not sufficient in many low- and middle-income countries to reach the goal of young children learning foundational skills. As a result, broadening the scope of education interventions beyond the school is a necessary step. This is where FH’s Education Care Groups come into play.

Care Groups consist of 10–15 community-based volunteer educators who meet regularly with project staff for training. They then facilitate behavior-change lessons to their neighbors in group settings and through household visits. Care Groups build caregiver capacity to support their children’s early child development and early-grade success.

FH staff involved in Care Groups for education have seen caregivers receive the information and use it in their own lives, in addition to the everyday lives of their children. One FH staff member said, “Before, they (caregivers) didn’t have time to think about their children’s studies. But since joining Care Groups, they are always trying to find time to help children.”

A staff member observed during a home visit that a father was carving a toy helicopter for his child. This shows that parents have learned the importance of play for child development and have found ways to make toys with locally available materials. FH staff is encouraged by this and say they “have hope that children will have their future with education. I think this program helps us to achieve our goal in every corner.”

Recent research by FH and Boston College’s School of Social Work has shown promising results of the impact of Care Groups on child learning outcomes. Researchers analyzed data sets from 10 countries that included data from a caregiver survey, as well as child learning outcomes measured by the International Development and Early Learning Assessment (IDELA) for children 3-6 yrs, and the Citizen Led Assessment (CLA) for children 7-15 yrs.

Researchers first compared two groups - caregivers who participated in Care Groups for education and those who did not, and correlated that with a caregiver behavior score, which was a sum score of multiple targeted learning support behaviors. These behaviors included:

-       Setting a specific space for study in the home

-       Reading with your child every day

-       Meeting with your child’s teachers

-       Singing songs and telling stories with your children

-       Playing with your children

-       Checking your children’s homework

-       Enrolling your children in school at the recommended age

The correlation analysis revealed a statistically significant positive association between participation in Care Groups and caregiver behavior scores.

Researchers also analyzed the learning outcomes of children of the caregivers who participated in Care Groups and found that these children had higher scores on IDELA and CLA than children of caregivers who did not participate. Additionally, researchers found stronger correlations between caregiver behavior scores and child learning outcomes, when compared to participation alone. 

In summary, caregivers who participate in community-based platforms that integrate social behavior change science, such as Care Groups, and adopt the early learning support behaviors have children who tend to score higher on learning assessments. Caregivers continue to be an important partner in the learning and development process.

Celebrating Teacher Appreciation Week with a Global Teacher and Teacher Coach Study

In honor of teacher appreciation week, let’s take a look at a fascinating global teacher and teacher coach study conducted by BEC member Teach for All. Below, Robbie Dean (Teach for All’s Global Director of Research, Evaluation & Network Learning and BEC’s Monitoring, Evaluation, Research, and Learning Co-Chair) shares about their first pilot effort to implement applied research to learn about the Teaching As Collective Leadership framework.

As a member of Teach For All’s research team, I support our network to explore an ambitious learning and research agenda on the impact of collective leadership development and how to accelerate the impact of teachers to develop classrooms that cultivate students as leaders of a better future. In partnership with the Jacobs’ Foundation, our team is working to build our global network’s internal research capability and experimenting with how we incorporate new forms of evidence generation and feedback into our approach to network learning, in addition to providing feedback on important learning questions associated with our Teaching As Collective Leadership (TACL) initiative, a framework for developing students as leaders that includes actionable and locally customizable guidance for teachers, teacher coaches, and program designers who are oriented towards this different view of the purpose of education.

The Global Teacher and Teacher Coach Study marks our first pilot effort to implement applied research to learn about the Teaching As Collective Leadership framework. The study data is from 48 teachers and 24 teacher coaches representing nine network partners across varied contexts. Our study investigates to what extent teachers develop mindsets aligned with collective leadership, and what kinds of classroom strategies they engage in as a result of exposure to the TACL model. In 2022, we tested two different delivery methods of the TACL framework: a series of workshops between teachers and their teacher coaches, and a comparison group of teachers and their teacher coaches in which we provided virtual instructions on how to use the TACL framework and resources. Both the intervention and comparison groups were designed to emulate common ways Teach For All offers learning experiences and insights on teacher development to our network partners.  

In both groups, we engaged with teachers and teacher coaches from across the Teach For All network to implement insights, mindsets, and instructional strategies in their classrooms based on the TACL framework. Recently, the study was featured in an episode of the BOLD Podcast, which highlighted the experiences of three network teachers: Aljawhara Al Athba from Teach For Qatar, Isaac Galindo from Enseña por México, and Ganiyat Muritala Wuraola from Teach For Nigeria. These stories provide a snapshot of the innovations that teachers and their teacher coaches are implementing in their classrooms to foster student leadership, including more concerted efforts to support students’ sense of belonging, a greater emphasis on strong teacher-student relationships, and strategies for social and emotional learning development. Hear more about their suggested strategies for other teachers in this podcast.

Our study report is helping our global network to learn more about the enabling conditions to support teachers to develop students as leaders. The following are a few major insights from both quantitative and qualitative data:

• Teacher mindsets are highly aligned with TACL already, and they also can strengthen over time: Overall, both teacher groups (our treatment and comparison group) tended to score high in terms of their mindsets—such as their beliefs in community as a source of power and the extent to which educational inequity is systemic—suggesting that teachers may have already had strong alignment with the beliefs and mindsets associated with TACL prior to engaging with the framework. Nevertheless, our analysis indicates some evidence that more frequent exposure to TACL concepts may in fact increase these mindsets over a short time period.

• Teacher coaches are key: Unilaterally, we heard from teachers that their teacher coaches were key in supporting them to apply the TACL framework. We also heard from teachers whose teacher coaches consistently applied the tools and resources and provided feedback, that they felt they had stronger knowledge of the framework and also noticed more shifts in their mindsets and changes in their actions. We hear this consistently in the stories shared by the teachers interviewed in the BOLD podcast. 

• Opportunities for teachers to share their experiences with other educators matter: Many teachers and their teacher coaches shared that they deeply appreciated and learned from discussing their experiences in the classroom with other teachers and teacher coaches from across the network, in particular those who participated in the intervention group. 

• There is strong evidence from the network that teachers are fostering environments to empathize and connect with their students: We’ve demonstrated evidence that teachers in our study quickly adopted strategies related to building strong, empathetic relationships with their students. We heard from teachers and coaches that they learned more about the importance of building meaningful relationships with students through the experience, and this came out clearly in interviews. 

We also received constructive feedback and challenging findings from our data and from the experiences of our teachers and teacher coaches in this study. For instance, we found fewer indications than expected of teacher strategies that emphasize student-driven learning and student autonomy in the classroom. This may be an area for further intervention to help teacher coaches and teachers create plans for more holistic approaches to employing multiple strategies in the classroom with equal emphasis. We also heard from teachers and teacher coaches that they would like to have more time to first internalize the model and make meaning of this framework, and additional time to implement and reflect on how the model works in their own classrooms to help strengthen student leadership. We are confident that this feedback and the iterative design process we are taking with the TACL team is helping us to strengthen the framework and resources, and providing our research team with new tools and opportunities to apply a new adaptive research and learning approach. 

The Global Teacher and Teacher Coach Study is just the second of three evidence generation/feedback activities around the Teaching As Collective Leadership Framework in partnership with Jacobs Foundation. In 2021, we preceded this study by a beta-testing of TACL tools. One important limitation of our study is that it did not aim to evaluate the potential effects of the model and interventions on students. For our future research, we will strive to determine whether these shifts in mindsets and teacher actions lead to improvements in holistic student learning outcomes by working with a network partner (Enseña por Colombia) to adapt and implement programmatic changes to its training and support of teachers in 2023 and 2024. We are excited to continue to learn and share our insights from our adaptive learning and research strategy with practitioners within and beyond our global network.