Educational AI Risks Overlooking Multilingual Learners
As artificial intelligence enters classrooms, experts warn that over 5 million multilingual learners risk being sidelined unless systems are intentionally designed for them from the start.
Artificial intelligence is rapidly transforming classrooms, changing how educators teach and students learn. However, as schools and edtech companies rush to integrate these technologies, a critical question arises: Are these platforms being built to support all students, including the more than 5 million multilingual learners (MLLs) in our schools? According to Crystal Gonzales, founder and executive director of the nonprofit English Learners Success Forum, the rush to adopt new tech risks repeating familiar mistakes.
Every technological system embeds assumptions about language, learning, and students within its training data. If that data lacks representation for MLLs and students with disabilities, or relies on educators untrained in supporting multilingual learners, the resulting software can be incomplete or biased. While simple translation features can help students access material, they do not automatically foster English proficiency, deep subject knowledge, or active participation.
Practical Ways to Integrate AI for MLLs
Real instructional design for multilingual learners must incorporate research-based practices rather than superficial add-ons. Gonzales highlights several specific, practical ways developers can use AI to strengthen instruction:
- Actionable Teacher Guidance: Drafting specific feedback prompts, targeted questions, and listening cues grounded in actual student work and objectives, avoiding generic scaffolds.
- Language Demands Analysis: Analyzing and drafting the language demands within a text or task, integrating language considerations from the outset.
- Sample Student Responses: Generating sample responses that demonstrate content understanding across various English proficiency levels, helping teachers recognize strong thinking despite developing language skills.
Without these safeguards, AI systems risk mistaking developing English proficiency for a lack of content knowledge. A student might fully grasp a scientific concept or complex problem-solving technique without possessing the advanced English skills needed to articulate it. Failing to separate language proficiency from subject knowledge distorts feedback and limits student opportunities.
Education has faced similar challenges before, notably during the rollout of the Common Core State Standards in the early 2010s. Many districts adopted materials built primarily for students in the middle, treating the needs of MLLs as secondary accommodations. Fixing those historical gaps proved difficult, and AI risks repeating those exact patterns at a much greater speed and scale.
What it means for developers
For developers, investors, and school districts, current decisions carry heavy consequences. Developers must integrate MLL expertise directly into core product teams, testing platforms continuously with multilingual students and their teachers throughout development rather than relying on brief post-launch checks. School districts need to vet tools rigorously before adoption, questioning whether solutions support both language acquisition and grade-level learning, whether they distinguish between English proficiency and subject-area knowledge, and if actual MLLs helped test them.
Developers looking to build robust, multilingual-aware applications can try top AI models cheaply through one API at https://apixoai.online. Designing for multilingual learners from the outset ultimately creates stronger, more effective educational systems for all students.
Source: Opinion: Don't Let AI Repeat Old Mistakes With Multilingual Learners — GovTech AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
