Investigating the Impact of AI-Supported Self-Coaching as a Professional Development Model for Embedded Instruction in Inclusive Early Childhood Settings
An AI self-coaching app can replace live experts and still push teacher fidelity and child skills way up.
01Research in Context
What this study did
Balikci (2026) tested an AI phone app that coaches preschool teachers while they teach. Teachers filmed short clips of embedded-instruction moments with children with autism. The app gave instant feedback: praise, fix-next-time notes, and a new goal.
The study used a multiple-baseline design across teachers. Each teacher moved through plan-act-reflect cycles on their own. No live coach visited the room.
What they found
Teacher fidelity jumped up fast and stayed high. Kids also gained the target skills tied to each lesson. Gains were large and lasted after the app stopped.
Every teacher reached the mastery line with only the AI prompts. One teacher said it felt like ‘a coach in my pocket.’
How this fits with other research
The result lines up with Castañe et al. (1993), the first study to show brief daily coaching lifts staff and child behavior. Balikci simply swaps the human coach for an algorithm.
Aguilar et al. (2023) used bug-in-ear telehealth to coach parents during Google Slides play. Balikci moves the same real-time audio idea into an AI loop for teachers.
Kunze et al. (2025) showed novice clinicians can coach parents on Zoom and still get strong child gains. Balikci goes one step further: it removes the human coach entirely.
Panganiban et al. (2022) trained teachers to embed JASPER play; gains were modest. Balikci’s AI feedback produced larger fidelity leaps, hinting that instant data-driven hints beat annual workshops.
Why it matters
You no longer need to drive across town to give feedback. Hand a teacher the AI app, teach the loop once, and fidelity rises on its own. Use it for embedded instruction, peer play, or transition routines. The tool frees you to design lessons while the app handles the daily coaching.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Film a 3-minute clip of a teacher doing embedded instruction, load it into a free AI feedback tool, and share the auto-generated next step.
02At a glance
03Original abstract
This study examined the effectiveness of an Artificial Intelligence (AI)-supported self-coaching system designed to improve preschool teachers’ implementation of embedded instruction (EI) for young children with autism in inclusive early childhood classrooms. Using a multiple-probe across participants single-case design with four teacher–child dyads, the study evaluated changes in teacher fidelity, child learning outcomes, maintenance, generalization, and teacher perceptions. Following baseline and an initial EI training, teachers engaged in weekly AI-supported self-coaching cycles that included planning, data entry, reflection, and AI-generated individualized feedback. Results demonstrated clear functional relations between the introduction of the AI-supported system and increases in teachers’ EI fidelity. All teachers reached high levels of accurate implementation, maintained their performance after AI supports were withdrawn, and generalized EI procedures to non-targeted routines. Correspondingly, children showed substantial improvements in unprompted correct responding on individualized goals, with gains sustained across maintenance and generalization probes. Social validity data indicated that teachers found both EI and AI-supported self-coaching highly acceptable, feasible, and helpful for guiding instructional decision-making. Findings provide promising initial evidence that AI-supported self-coaching can serve as a scalable, cost-effective professional development approach that strengthens teacher practice and enhances learning outcomes for young children with autism in inclusive preschool settings.
Behavioral Sciences, 2026 · doi:10.3390/bs16010140