Attributional patterns toward students with and without learning disabilities: Artificial intelligence models vs. trainee teachers.
AI stays calmer and kinder toward students with LD yet grades feedback more harshly—blend AI optimism with your human touch.
01Research in Context
What this study did
The team asked two groups to read short stories about students. One group was new teachers. The other group was large-language AI models like ChatGPT.
Each story told of a student who failed a task. Some students had a learning disability label. Some did not.
After each story, both groups rated how frustrated, sympathetic, and hopeful they felt toward the student. They also judged the quality of teacher feedback the student had received.
What they found
The AI models felt less frustration and more sympathy than the trainee teachers. They were also more hopeful that the student would improve.
Surprisingly, the AI models called the same teacher feedback "worse" than the humans did.
The pattern held for both LD and non-LD students. The machines were simply softer on kids and tougher on feedback.
How this fits with other research
Patton et al. (2020) found that real teachers give more negative comments to Black students even when behavior is equal. Inbar et al. now show that AI, by contrast, dishes out less blame to students yet harsher scores to feedback. Together the papers warn us that human feedback can carry hidden bias, while AI feedback may need human warmth added.
Kiliç Tülü et al. (2026) looked at 733 Israeli elementary students and learned that grades and behavior, not the LD label, shaped teacher views of social skills. Inbar et al. echo this: both studies say teacher expectations hinge more on performance than on the disability tag itself.
Poppes et al. (2016) tried a short workshop to shift staff explanations of challenging behavior and saw almost no change. Inbar et al. did not train anyone; they simply swapped in AI minds and saw instant attitude shifts. The comparison hints that technology, not a one-day lecture, may move attributions faster.
Why it matters
If you write IEP goals or give student feedback, this study is a wake-up call. AI tools can help you stay optimistic and reduce frustration language, but they may also word feedback too harshly. Run AI drafts through your human filter before sharing. Aim for AI-level hope plus teacher-level tact.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Next time you use AI to write feedback, paste the draft into your doc, soften any harsh phrases, then keep the hopeful tone before sending it home.
02At a glance
03Original abstract
This study explored differences in the attributional patterns of four advanced artificial intelligence (AI) Large Language Models (LLMs): ChatGPT3.5, ChatGPT4, Claude, and Gemini) by focusing on feedback, frustration, sympathy, and expectations of future failure among students with and without learning disabilities (LD). These findings were compared with responses from a sample of Australian and Chinese trainee teachers, comprising individuals nearing qualification with varied demographic and educational backgrounds. Eight vignettes depicting students with varying abilities and efforts were evaluated by the LLMs ten times each, resulting in 320 evaluations, with trainee teachers providing comparable ratings. For LD students, the LLMs exhibited lower frustration and higher sympathy than trainee teachers, while for non-LD students, LLMs similarly showed lower frustration, with ChatGPT3.5 aligning closely with Chinese teachers and ChatGPT4 demonstrating more sympathy than both teacher groups. Notably, LLMs expressed lower expectations of future academic failure for both LD and non-LD students compared to trainee teachers. Regarding feedback, the findings reflect ratings of the qualitative nature of feedback LLMs and teachers would provide, rather than actual feedback text. The LLMs, particularly ChatGPT3.5 and Gemini, were rated as providing more negative feedback than trainee teachers, while ChatGPT4 provided more positive ratings for both LD and non-LD students, aligning with Chinese teachers in some cases. These findings suggest that LLMs may promote a positive and inclusive outlook for LD students by exhibiting lower judgmental tendencies and higher optimism. However, their tendency to rate feedback more negatively than trainee teachers highlights the need to recalibrate AI tools to better align with cultural and emotional nuances.
Research in developmental disabilities, 2025 · doi:10.1016/j.ridd.2025.104970