The promise of artificial intelligence in education has grown rapidly, and a new working paper from the Organisation for Economic Co‑operation and Development (OECD) titled “Leveraging Artificial Intelligence to Support Students with Special Education Needs” provides a thoughtful overview of how AI can support learners—but with major caveats.
At its core, the report argues that AI tools which adapt instruction, generate accessible content and provide support tailored to individual learners have real potential in special education, therapy and inclusive classrooms. For example, an AI system might generate simplified reading passages for students with dyslexia, create visual supports or scaffolds for students with language delays, or adapt pace and format for students with attention or processing challenges.
For therapists and special educators, this means opportunities to innovate. Instead of manually creating multiple versions of a lesson or communication script, generative AI can support you by producing varied, adapted material quickly. A speech therapist working with bilingual children might use an AI tool to produce scaffolded materials across languages; an occupational therapist might generate tactile-task instructions or interactive supports that match a student’s profile.
However, the OECD report also emphasises that equity, access and human-centred design must accompany these possibilities. AI tools often rely on data trained on typical learners, not those with rare communication profiles or disabilities. Bias, representation gaps and access inequities (such as device availability or internet access) are real obstacles.
In practice, you might pilot an AI-driven tool in one classroom or one caseload, with clear parameters: what are the outcomes? How did students engage? Did the tool genuinely reduce the manual load? Did it increase learner autonomy or scaffold more meaningful interaction? Collecting student and family feedback, documenting changes in engagement, and reflecting on how the tool leveraged or altered human support is key.
Inclusive AI also demands that you remain the designer of the environment, not the tool. For example, when generating supports for a student with autism and a co-occurring language disorder, you might ask: did the AI produce appropriate language level? Did it respect cultural/language context? Do hardware/internet constraints limit access at home or in school? These reflections help avoid inadvertently widening the gap for students who may have fewer resources.
From a professional development perspective, this is also a moment to embed AI literacy into your practice. As learners engage with AI tools, ask how their interaction changes: Are they more independent? Did scaffolded tools reduce frustration? Are they using supports in ways you did not anticipate? Part of your emerging role may be to monitor and guide how students interact with AI as part of the learning ecology.
If you’re exploring inclusive AI, consider creating a small pilot plan: select one AI-tool, one student group, one outcome metric (e.g., reading comprehension, self-regulation, communication initiation). Run a baseline, implement the tool, reflect weekly, and refine prompts or scaffolded supports. Share findings with colleagues—these insights are vital for building sustainable AI-assisted practice.
Suggested Reading: