Harnessing AI for More Equitable Outcomes
The field of artificial intelligence (AI) is rapidly evolving, with the power to disrupt and transform the social and education sectors. Recently, the National Science Foundation (NSF) announced the creation of an AI Institute for Exceptional Education, focused on “developing advanced AI technologies to scale availability of speech language pathology services.” This exciting development illustrates the power AI has to support practitioners and students with new tools for learning.
Many of the practitioners, funders, and policymakers we’ve spoken to are unsure of what the near future holds—they are excited by opportunities to leverage AI to increase efficiency and scale innovation, but wary of the potential for bias or misuse to lead to unintended consequences. At Project Evident, we believe that thoughtful and responsible AI applications can lead to stronger and more equitable student outcomes by supporting practitioner decision making, accelerating R&D and evidence building, and strengthening equity and accessibility.
Supporting practitioner decision making
One of the most exciting opportunities we see for AI is for it to unlock the power of data and evidence to support practitioners in their work. First Place for Youth, a nonprofit that helps foster youth build the skills they need to make a successful transition to self-sufficiency and responsible adulthood, uses “precision analytics” and machine learning to build predictive, prescriptive, and evaluation models to generate actionable evidence for their front-line staff. This process uses program administration and case assessment data to learn from differences in outcomes among youth, helping FPFY’s staff better understand what has worked for specific participant populations in the past and what services and supports are most likely to lead to success. At the education nonprofit Digital Promise, their team is using AI to power a recommendation engine that will help connect educators to relevant professional learning opportunities. The recommendation engine will allow them to scale their work of building the capacity of teachers to better serve diverse learners, especially those who are underserved. Both examples illustrate how AI algorithms can formulate meaningful insights from data and evidence to provide critical decision-making support and feedback to practitioners.
Accelerating research and development (R&D) and evidence building
For too long, research in the social and education sectors has prioritized large-scale, multi-year studies that produce thumbs-up/thumbs-down results that are of little use for program improvement. By embracing an R&D approach focused on continuous testing and learning, practitioners can enable more timely and relevant improvements. At the early learning platform Noggin, R&D is a key component of their content production pipeline. A cross-disciplinary team of content developers, instructional design experts, and research scientists have developed a process for rapid-cycle research that allows them to continuously improve content as it is being developed and to quickly and economically test for evidence of learning impact. In another example, the job training program Year Up employs rapid-cycle randomized trials to compare alternative academic support strategies and to guide program improvement; multiple studies have found significant wage increases among program graduates. AI can enable more organizations to build their capacity for R&D by leveraging existing data, recognizing patterns, and unlocking real-time insights to support continuous learning and improvement.
Strengthening equity and accessibility
Many practitioners are skeptical of AI, with good reason. AI is based on historical data; given our nation’s history of marginalization, these biases permeate the tools themselves. Practitioners also worry about opportunities for misuse, including cheating and job displacement. However, with thoughtful design and implementation, we have an opportunity to leverage AI to advance equity. Similarly, by increasing our awareness of misuse, we can adapt our practices and accelerate our understanding of how to harness AI-powered approaches for good. Equal Opportunity Schools (EOS), a nonprofit focused on increasing equity in public education, identifies students of color and students from low-income backgrounds with the potential to succeed in Advanced Placement and International Baccalaureate courses. Their team is now exploring using AI to scale their program and to increase the accuracy and efficiency of their student identification process. While there was some initial fear that by relying on AI they would lose the “human” element of their work, they’ve come to understand that by reducing the labor intensiveness required for partnership managers to manually sift through student and school data, automating those systems would in fact free up staff to do more direct work with individual students while also allowing the organization to broaden their overall reach.
Another clear way these approaches can support equity in education is by increasing accessibility, as the new AI Institute for Exceptional Education aims to do. The National Science Foundation shared, “The need for speech and language services has been exacerbated during the COVID-19 pandemic due to a widening gap in services available to children. The AI Institute for Exceptional Education aims to close this gap by developing advanced AI technologies to scale availability of speech language pathology services so every child in need has access.” AI-powered assistive technologies can provide students with disabilities with a wider range of tailored educational materials, services, and experiences. For example, AI-powered speech recognition can help students who are blind or have visual impairments to access text-based materials, while AI-powered text-to-speech can help students who are deaf or hard of hearing to access audio-based materials. AI algorithms can also analyze a student’s performance and inform personalized instructional decisions to support learning, particularly for students with disabilities who may require more individualized support.
AI has enormous potential to support stronger and more equitable outcomes in education. Unfortunately, financial and capacity restraints mean that the work required to design and implement AI solutions is currently out of reach for too many nonprofits and school districts. As Digital Promise’s Director of Information Technology Diane Doersch shared, “I always dreamed of stuff like this, but as an educational entity you don’t have the money to be able to imagine these types of things and then make it work.” A small but growing number of funders, including Schmidt Futures, the Walton Family Foundation, and the Patrick J. McGovern Foundation, are working to change that by supporting equitable artificial intelligence and R&D efforts. We also applaud the Institute for Education Sciences, the Alliance for Learning Innovation, InnovateEDU, and others who are moving the field forward. The creation of the AI Institute for Exceptional Education is an exciting development that highlights the power of investing in AI technologies to test and scale innovations that are deeply relevant to students and families. At Project Evident, our goal is to advance a next generation of R&D and AI practice, where use cases like the ones highlighted in this piece are the norm rather than the exception, and where AI-powered solutions can provide millions of students with more effective and equitable education opportunities.