Christos A. Makridis and Goldy Brown: The potential of AI systems to improve teacher effectiveness In Spokane Public Schools

By Christos A. Makridis and Goldy Brown

The United States K-12 education system has faced challenges for years, but has faced even greater headwinds recently following the pervasiveness of school closures and the resulting effects on student mental health and learning outcomes. Student test scores in math and reading fell to their lowest levels in 2022, according to the National Assessment of Educational Progress. These deteriorating outcomes warrant effective instruction from teachers across classrooms.

This year, Spokane Public Schools announced that it is pioneering a novel approach to evaluating and improving teacher effectiveness using AI systems. While AI is sometimes thought of as displacing jobs, it can also augment our productivity and learning. And as this school district in Spokane is exploring, AI systems can potentially help lower-performing teachers improve their quality of instruction at scale and embed greater consistency into teaching nationwide.

School districts often struggle with limited resources to provide continuous, quality training for their teachers, and bureaucratic impediments make removing ineffective teachers an arduous process. As a result, many students suffer under the instruction of teachers who, despite their best intentions, are ill-equipped to meet their educational needs. A large body of empirical research, in part led up by professor Eric Hanushek at the Hoover Institution, has pointed out that teacher quality is the single greatest impediment to learning outcomes.

The recent advances in large language models, such as Bard and ChatGPT, highlight the ways that AI can improve training and assessment of teachers at scale without having to involve principals and other training professionals for each individualized case. In particular, AI-powered platforms can provide a personalized, data-driven approach to teacher training.

By analyzing classroom data and building statistical models that predict learning outcomes as a function of teacher characteristics and inputs, these systems can offer real-time feedback and guidance, addressing teachers’ specific areas of weakness and offering them ways to improve. For example, if a teacher consistently struggles with engaging students or explaining complex topics, the AI could provide tailored strategies and methods to improve in these areas.

Moreover, AI-based coaching systems offer scalability and efficiency that traditional teacher training programs cannot match. Such a system can serve numerous teachers simultaneously, providing continuous support and learning opportunities. This continuous feedback loop would allow teachers to refine their skills constantly and adapt their teaching styles to their students’ evolving needs. Furthermore, AI systems would avoid putting further strain on the educational system that has already been stretched thin post-COVID.

While the potential benefits of AI in teacher coaching are vast, successfully implementing such a system requires careful consideration. An essential aspect of managing these AI systems is ensuring they are ethically used and respect teachers’ and students’ privacy. Confidentiality of data is paramount, and AI systems must be designed and regulated to ensure they comply with laws and ethical guidelines pertaining to data protection.

For example, our recent book, “The Economics of Equity in K-12 Education: A Post-Pandemic Policy Handbook for Closing the Opportunity Gap and Using Education to Improve the American Economy,” prominently features recommendations by professor Ryan Baker and University of Pennsylvania, emphasizing that the use of AI in education will require data sharing between schools and vendors using the latest advances in cryptography, like zero knowledge proofs, to secure sensitive information.

Additionally, AI systems need regular fine-tuning to remain effective and relevant. This process would involve an ongoing cycle of feedback from teachers, AI developers and education experts to ensure the AI evolves in line with the changing dynamics of classrooms and the educational landscape. For instance, updates in curriculum, pedagogical strategies and teaching methods should be reflected in the AI’s feedback and coaching suggestions.

Ultimately, AI is a tool, rather than a replacement for human connection and judgment: Decisions must remain with the educators and administrators. AI can provide data-driven insights and recommendations, but it’s the teachers and administrators who will interpret this information in the context of their unique classroom environments and make the final decisions.

The pioneering work by Spokane Public Schools represents a novel attempt to solve the longstanding challenge of the deterioration in student learning outcomes driven, at least in large part, by the decline in teacher quality and absence of incentives. With careful management and continuous refinement, AI systems could revolutionize teacher coaching, significantly improving the quality of K-12 education across the nation.

While challenges remain, the path forward shows immense promise, offering hope to educators and students alike.

Goldy Brown III is an associate professor in the Graduate School of Education at Whitworth University in Spokane . He is also the director of Whitworth University’s Educational Administration Program. He was a school principal for almost a decade. Schools that he administered earned four state recognition awards for closing the achievement gap between low-income and affluent students.

Christos A. Makridis is a research affiliate at Stanford University’s Digital Economy Lab and COO/co-founder of Living Opera, a multimedia startup focused on empowering and educating performing artists. He holds doctorates and masters degrees in economics and management science and engineering from Stanford University.