Executive functioning is a term applied to a wide range of functional self-management skills that are challenging for students with disabilities. From childhood to adulthood, skills such as planning, prioritizing, time management, organization, working memory, metacognition, focus, flexibility, task initiation, perseverance, inhibition and goal-setting can be impacted by a person’s disability. Students who demonstrate executive functioning deficits may have one or more of a wide range of disabilities, including: Attention Deficit Hyperactivity Disorder (ADHD), Autism Spectrum Disorder (ASD), Intellectual Disabilities (ID), Specific Learning Disabilities (SLD), Other Health Impairments (OHI), etc.. A need for the development of executive functioning skills is critical regardless of whether students with disabilities are learning basic vocational tasks or are working on their doctoral theses. In addition to academics, these skills support social interactions, emotional regulation, and independent living skills (Riley, Jermone-Raja, & Chakravarthi, 2025).

Finding a way to support the executive functioning skills of students with disabilities is both personally and professionally important to me. My husband and I have five children, including a son with Down Syndrome, a daughter with high functioning Autism and two children with ADHD. All five of them, even the youngest who does not have a disability, would benefit from developing skills in self-control, planning and flexibility. For the last decade, I have also been a Learning Behavior Specialist (special education teacher) in the Chicago suburbs. Currently, I teach high school students who are primarily college bound. They have the academic knowledge to do well, but a lack of executive functioning skills can present significant obstacles to post-secondary success. My students may hesitate to begin large projects, because they don’t know where to start. Sometimes, they fall behind on assignments due to a lack of prioritization, focus, and time management skills. These skills are needed to be effective students, employees and friends.

In practice, an example of applying executive functioning skills in an academic setting would be: focusing on a class lecture, identifying key information, using working memory to retain the information while formatting it into organized notes, exhibiting self-control to avoid playing a game or other off-task behavior when given class time to begin an essay, organizing the essay topic into an outline, planning to reach the larger essay due date by completing a paragraph each day, persevering through the writing process, demonstrating flexibility when ideas lead in a new direction, managing their time between multiple classes and other obligations, demonstrating task initiation to work towards their academic goals each night, and demonstrating metacognition during the revision process. An example of applying these same skills in a social setting would be for a student to observe a group of other students playing after school, focus on figuring out the organization of their game, initiating taking a turn, being flexible if asked to wait, persevering while earning how to play, monitoring the time before they had to leave to go home, and planning when to get together again.

Recently, an almost dizzying quantity of AI supports and assistive technology applications have emerged to support students. Grammarly, Read & Write, speech-to-text, text-to-speech, closed captioning, ChatGPT, Magic School, OrbitNote, Notability, Google Suite, virtual assistants, websites, and countless chromebook extensions/applications, etc. threaten to overwhelm students and educators alike through sheer volume of choices. Indeed, identifying, analyzing and matching the technology to the individual needs of students becomes of importance for all stakeholders: technology developers, educators, individuals with disabilities, and their families. As AI and assistive technology tools are constantly changing or being updated, ongoing training and collaboration is essential for everyone involved. With K-12 students it is particularly important to include students’ families in the selection and training process as acceptance of the technology at home increases student use (Nieves, 2021).

Ironically, the process of identifying needs, researching available resources, matching the correct resource to the need, learning how to use the tool effectively, analyzing the success of the tool and making adjustments requires executive functioning skills. This presents additional challenges as students graduate from high school and lose the familiar support of their school districts. Therefore, it is particularly important in post-secondary settings that students are aware of both people and technologies that are available to assist with problem solving any obstacles which may arise in a new environment. In fact, one current research article focused on the development of a virtual assistant to support students in the process of disclosing their disabilities to their university in an accessible manner (Lister et al., 2021).

Although initially, Executive Functioning and AI: Support for Students with Disabilities seemed like it might be a prohibitively narrow topic, there has been a recent explosion of research in this area. In addition to the aforementioned virtual assistant, there are multiple articles about the use of smartwatches to support students with intellectual disabilities. Several resources cover the use of ChatGPT to support executive functioning in written expression and other areas which would be tremendously helpful to my college bound students. A 2025 book compilation explores the use of AI specifically in Special Education. Other research investigates how AI can evaluate student engagement or emotional states. While critics discuss the ethics of viewing privacy concerns as a necessary trade-off for effective accommodations, other research advocates for the participation of individuals with disabilities in technology development. There is so much information that it is difficult to do more than scratch the surface.

The teaching of executive functioning skills can be approached through the lens of multiple theoretical viewpoints. Behaviorists view the action or inaction of executive functioning skills like task-initiation, planning, organization, time management and others in terms of positive or negative reinforcements connected to students’ actions or inactions. B.F. Skinner emphasizes the importance of immediate, consistent feedback (Kalantzis & Cope, 2024). Although, he also acknowledges the student’s capacity to provide their own intermediate reinforcers outside of the classroom through checking their own work or by recognizing their own progress (Kalantzis & Cope, 2024). Within the classroom itself, Skinner acknowledges the inherent difficulties of scaling immediate, consistent feedback from the teacher to the student as instruction moves from smaller groups to large lecture halls (Kalantzis & Cope, 2024). In modern online learning, Skinner’s thoughts regarding executive functioning become quite apropos. In the 2024, online text, New Learning, by Kalantzis & Cope, Skinner’s Behaviorism, paragraph five, references B.F.Skinner’s own words from Technology of Teaching, “Education has never taught the self-management of motivation very effectively. It has seldom tried. But techniques become available as soon as the problem is understood …” One of the techniques which Skinner applies to this problem is that of machine learning (Kalantzis & Cope, 2024). Given this context, it seems likely that incorporating the scalable, 1:1, immediate feedback potentials of AI to a modern classroom would be a path that behaviorists would endorse. Interestingly, in the TedTalk linked below, Open AI Co Founder Greg Brockman, shares that the manner in which they taught ChatGPT to learn was with a behaviorist approach. In this clip, Mr. Brockman shares that they followed Alan Turing’s ideas for teaching a machine to learn. They treated it like a child and provided it with feedback in the manner espoused by Behaviorism (Brockman, 2023). Surely, Skinner would approve of the full circle this implies.

The Inside Story of ChatGPT's Astonishing Potential, watch 5:29-8:42

The Montessori approach to teaching executive functioning skills is quite different. Rather than the teacher rewarding external behaviors, the environment is carefully designed to engage and support the development of self-control in order to support internal motivation (Lillard, 2017). In Montessori, the Mind Behind the Genius, by Angeline Lillard, Maria Montessori is quoted as saying,

In the old school, one goes from the outside and then evidently it is the teacher who has the greatest role: the teacher must first work to achieve order and then she must put in the child’s mind those things which she wishes him to learn. Here, on the other hand, progress comes from within the child. It is the child who begins to exercise himself; he constructs his own conduct, so that this order becomes a phenomenon of development. (1997, p. 154)

(Lillard, 2017, page 105)

This typifies a Montessori approach to Executive Functioning which classes all of these skills underneath 3 primary headings: inhibition (self-control), working memory, and shifting (flexibility) (Lillard, 2017). Executive functioning is recognized as being predictive of the student’s future life success, outstripping other measurable factors such as intelligence, parent education, or student education (Lillard, 2017). Lillard also points out that this connects with Brain Developmentalism in that the prefrontal cortex continues to mature well into a student’s 20s, and that this is reflected in Montessori’s four stages of development which are outlined in the first chart below. Montessori’s four planes of development are the: Absorbent Mind, Reasoning Mind, Humanist Mind, and Specialist Mind (Lillard, 2017). These stages are also identified with the typical ages associated with each from birth to twenty-four years of age (Lillard, 2017). The second chart is interesting in that it shows how Montessori’s four stages overlap with those of other educational theorists.

Maria Montessori's Four Planes of Development, https://actonacademycolumbus.com/blog/Montessori-stages-development-infancy

Montessori stages of development paired with other theorists, https://www.montessoriphysicaleducation.com/new-blog/montessoris-planes-of-development-and-jung-piaget-vygotsky-bandura-froebel-brunner-kohlberg-dewey-and-may2024/6/3

Regarding the intersection of executive functioning with disability, Lillard notes that ADHD in particular is associated with low executive functioning skill levels and reflects that self-regulation is necessary to support positive social behaviors. In fact, research has been shown that focus is linked to happiness which in turn may lead to positive social behaviors. Lillard also notes that this same research by Killingsworth and Gilbert in 2010 also noted that mind wandering can lead to unhappiness (not the reverse). Montessori offers specific ways to improve executive functioning skills, specifically by increasing mindfulness which is defined as “paying attention to what is happening in the present moment” (Lillard, 2017, page 112). In young children, this can be done through simple games such as Simon Says or the Silence which requires students to be very still and listen attentively for a short time (Lillard, 2017). Teaching students meditation and other mindfulness techniques leads to an increase in attention even after the period of mindfulness (Lillard, 2017). As part of its educational structure, Montessori recommends a three hour work cycle (Lillard, 2017). Lillard notes that research demonstrates that interruptions to a student’s concentration can lead to self-imposed distractions. Lillard references studies in which adults were found to take an average of 23 minutes to return to their original task after being interrupted and that 40% of the time adults will not even return to their original task (Lillard, 2017). This has significant implications for students working with technology. Montessori methods note how television and screen time use in general can be detrimental to student focus (Lillard, 2017).

However, AI and other assistive technologies can also be used to embed mindfulness meditation into student routines or to block potential distractions. Perhaps Montessori methods for carefully preparing the physical environment according to student readiness, could be applied to the virtual environment. Just as physical Montessori classrooms are designed to optimize choice and student initiative, perhaps this same deliberate practice could be used to prepare our virtual classrooms to support all students' executive functioning.

Designing learning environments to automatically incorporate the needs of all students is an important component of Universal Design for Learning. In Assistive Technologies for Differently Abled Students, Chapter 6, Ensuring Technology Integration in the Classroom Leads to Increased Accessibility: Using UDL as a Lens, by Frederic Fovet, 2022, there is an excellent description of the three key tenets of UDL.

UDL uses three simple principles in order to support this design process among educators: multiple means of representation (offering optimal flexibility in the ways information is provided to the learner), multiple means of action and expression (offering optimal flexibility in the ways the learner offers content and contributions, participates in class, and completes assignment), and multiple means of engagement (integrating optimal flexibility in the ways the learner can create an affective connection between their context and their lived experiences and the content of learning and teaching) (Fornauf & Erickson, 2020).

(Fovet, 2022)

Because putting these principles into practice requires an inherently multimodal form of education for all learners, there is a natural embrace of assistive technology including artificial intelligence (Fovet, 2022). In traditional classroom settings, there has been evidence that students with disabilities have hesitated to use their assistive technology accommodations because they perceive that it marks them as different or less able than their peers (Fovet, 2022). This appears to be the result of the medical model of disability, in which only students with IEP accommodations were granted access to costly assistive technology interventions (Fovet, 2022). Even though rejecting their accommodations put students with disabilities at a disadvantage, students often viewed the trade off for a possibly increased social standing to be worth it (Fovet, 2022; Nieves, 2021). By contrast, UDL manages to destigmatize the use of assistive technology by thoughtfully making all materials accessible to everyone from the beginning (Fovet, 2022; Nieves, 2021). UDL also has shown to be more likely to utilize accessibility features which are already built into all student devices to support students with and without disabilities (Nieves, 2021).. According to cast.org, the stated goal of UDL is learner agency which is defined by designing learning environments that are: purposeful, reflective, resourceful, authentic, strategic and action oriented. In regards to executive functioning, the sixth framework is often referred to as addressing those skills, however, the fifth, eighth and ninth frameworks listed below also are included under the umbrella of executive functioning (Cast, 2024). The three principles including all guidelines and checkpoints are pictured below (Cast, 2024).

The Universal Design for Learning Guidelines, https://udlguidelines.cast.org/

Within UDL there are three levels of assistive technology: low tech, mid tech, and high tech (Delisio & Bukaty, 2019). Low tech might include the use of pencil grips or white boards, mid tech might include audio books, computer extensions or software applications while chromebooks, smartboards and AI fall under high tech (Delisio & Bukaty, 2019). AI and other assistive technologies can bridge many of the gaps or barriers that students may otherwise experience through: speech to text, text to speech, modification of the reading level of passages, AI executive functioning support for the organization of ideas/writing, digital and other AT supported note taking options, guided notes, use of images/video/audio, closed captioning, interactive online texts, etc. is intentionally made available to all students (Delisio & Bukaty, 2019). Therefore, the needs of students with disabilities are naturally incorporated into the planning and resources of curriculum and lessons (Delisio & Bukaty, 2019). In the UDL classroom all three levels of assistive technology are often present and working together.

Each of these three theories has its critics. Behaviorism is faulted for not giving enough merit to metacognition and intrinsic motivation instead limiting human potential to what is predictable and trainable. “As an educational philosophy, stimulus and response seems mechanical and manipulative,” (Kalantzis & Cope, 2024, para. 14). Brain developmentalism and Constructivism are both critical of a strictly Behaviorism approach (Kalantzis & Cope, 2024). These critics would add that students have to be ready to learn the material presented, and that teachers should create the conditions to meet students where they are (Kalantzis & Cope, 2024). One criticism of Montessori methods is that they can be difficult to scale and are not as naturally inclusive of technological advances (Lillard, 2017). Other critics of Montessori methods point out that students may not always choose what is best for them and sometimes adults should lead them in specific directions (Kalantzis & Cope, 2024). Universal Design for Learning incorporates ideas from both Behaviorism and Montessori while allowing for natural integration of technological advances with the low tech or no tech tools listed previously (Delisio & Bukaty, 2019). However UDL also has come under review. One 2024 systematic literature review argues that, “Despite UDL having entered its 4th decade of development and research, there have been ongoing critiques of UDL for lacking clarity in definition, challenges with implementation, and insufficient evidence of effectiveness,” (Zhang et al., 2024, p. 1). This review found that the 9 UDL checkpoints are applied unevenly across research studies with most attention being given to the Representation principle followed by Action & Expression, then Engagement (Zhang et al., 2024). Zhang and their colleagues recommend that future research specify UDL checkpoints more clearly and establish a deliberately systematic implementation of UDL theory (Zhang et al., 2024). Another problem can be a lack of research clarity when multiple checkpoints claim to be covered by the same activity (Zhang et al., 2024). In the chart below, the literature review is broken down by its checkpoints into 101 studies aligned with the UDLprinciple of Representation, 68 with Action & Expression, and 59 with Engagement (Zhang et al., 2024).

Chart listing UDL studies sorted by Engagement, Representation, Action & Expression: https://doi.org/10.1007/s10648-024-09860-7

Regarding the practical integration of AI technologies into UDL lessons, teachers tend to be concerned that in order to implement UDL they need professional development training as well as additional planning time up front. One article which addresses some of these concerns is Technology Integration Among Special Education Certified Co-Teachers in the Inclusion Classroom, by Kathryn Nieves Licwinko. More than three hundred special education co-teachers participated. The teachers were from across the United States and represented forty-six states from grade levels pre-kindergarten through grade twelve (Licwinko, 2024). Researchers conducted semi-structured interviews to answer the following questions:

RQ1. What are the technology integration attitudes and practices of special education co-teachers?

RQ2. What are the lived experiences of special education co-teachers in integrating technology into the inclusion setting?

RQ3. In what ways does the qualitative interview data with special education teachers explain the quantitative data about their frequency of and attitudes toward technology integration in the inclusion setting?

(Licwinko, 2024, p. 365)

The results reflected that co-teaching teams of paired general education and special education teachers are most common at grades six through twelve (Licwinko, 2024). The author of this paper recommended that a Technological Pedagogical Content Knowledge Framework (TPACK) could be paired with UDL to create inclusive learning environments by leveraging the knowledge of both teachers in planning, instructing and assessing (Licwinko, 2024). The image below uses a Venn Diagram format to illustrate the overlapping skills of co-teaching teams (Licwinko, 2024). In order to create TPACK, content knowledge of what is being taught, pedagogical knowledge of the specific needs of students, and teachers’ knowledge of technology must be applied simultaneously by both the general education and the special education teachers working in concert (Licwinko, 2024).

Technological Pedagogical Content Knowledge Framework, https://doi.org/10.1177/01626434241227170

The author concluded that co-teachers would like more technological professional development, more time for planning with their co-teacher during the school day, and that student needs are driving technology integration (Licwinko, 2024). When schools prioritize planning time for co-teachers and professional learning communities the opportunity to deliberately design UDL curricula and lessons increases. Another item of interest is that more than half of the special education co-teacher had only been working with their general education counterparts for less than two years with less than a quarter of participants having worked with the same partner for more than four years (Licwinko, 2024).. This lack of experienced pairs may also be impacting the quality of technology integration (Licwinko, 2024).

Two studies were reviewed that looked at ChatGPT specifically. Using ChatGPT to Assist Writers with Planning Writing Tasks in an EFL Classroom, by You, Lin, Han, Chu and Dai explored the impacts of 26 EFL university students in Taiwan on shaping their writing in English. An emphasis was placed on the metacognition awareness and critical thinking skills of the students during the writing process (You et al., 2024). Students wrote two papers using AI as a tool (You et al., 2024). However, the first paper on Aromatherapy was written without any formal training on how to use AI and then a paper on 168 Intermittent Fasting was written after students were given explicit instruction around using ChatGPT for support (You et al., 2024). Parameters included questionnaires at the beginning and end of the semester (You et al., 2024). Students composed their writing on their computers in a physical classroom and were instructed to copy and paste all conversations with ChatGPT into a separate document for analysis (You et al., 2024). Students were trained how to create a semantic map to visually structure their writing planning and to monitor their metacognitive and critical thinking by asking themselves the following questions:

1. What are the task demands? What does the writing task require me to do?

2. What do I know about the topic? What else do I need to know to complete the task?

3. What stance should I take? What information do I need to form my arguments and support them?

Questions during the second part included:

1. What do I want ChatGPT to do?

2. Are the responses generated by ChatGPT relevant to my questions? If not, how can I rephrase them?

3. What further questions can I ask ChatGPT based on the previous responses?

4. Do I have enough information through the interactions with ChatGPT? If not, what are other possible sources of information?

(You et al., 2024, pp 243-244)

The writing process was broken down into the six categories listed in the table below (You et al., 2024). The blue columns represent how students chose to use AI before their training and the red indicates how often students wrote AI prompts in the category after AI training (You et al., 2024). The results demonstrate a shift from utilizing AI support primarily in the early stages of planning to feedback during the argument and supporting detail phases (You et al., 2024).

The Percentage of Each Category of Prompt in Both Writing Tasks, https://drive.google.com/file/d/114A3EMWd_l7kUjrTVTTijvkeG5NMXPzB/view?usp=sharing

The researchers posit that because students were implicitly taught to use semantic maps and metacognition, critical thinking skills, they may have been more goal focused on developing arguments and supporting details when using ChatGPT (You et al., 2024). The lack of use of AI support on developing rhetoric samples or finding evidence may be because students recognized the limitations of ChatGPT in documenting its sources or may have been because students knew that their interactions with AI were being monitored (You et al., 2024).

The results of the questionnaires revealed that 90% of students felt that working with ChatGPT supported the planning phase of their writing, 40% of participants indicated that they had difficulty rephrasing their prompts when ChatGPT did not produce the desired result. It is indicated that this may have been in part due to their limited English proficiency (You et al., 2024). In response to the results of their research, the authors propose a four part model for using ChatGPT (You et al., 2024). “The model has four components: ChatGPT as a tool, a monitoring mechanism consisting of critical thinking and metacognitive monitoring, a factor-controlled generator of prompts, and the planning process” (You et al., 2024, p. 254).

ChatGPT as a Universal Design for Learning Tool Supporting College Students with Disabilities, by Sandra Ayala challenges colleges to reassess accessibility. Ayala points out a remarkable array of capabilities such as: summarizing text, rephrasing in simpler language, adjust to multiple comprehension levels, incorporating text to speech, engage in dialogue to support conversational skills, assist with time management, planning, written expression, and that the AI can evolve over time to provide tailored support specific to individual needs. Ayala cautions that students will need training that includes how ChatGPT fits into the UDL framework as well as guidance and boundaries for appropriate usage, there is clear excitement for the wealth of possibilities. As the author states, “In essence, while ChatGPT serves as a multifunctional tool for all students, it functions as a lifeline for students with disabilities, offering them a degree of independence and self-determination in their educational journey that they might not otherwise have (Mazzotti et al., 2023)” (Ayala, 2023), p 32). This article also provides several pages worth of support topics, rationales and sample student prompts that are truly helpful and included in the sample below (Ayala, 2023).

ChatGPT Support Topics, Rationales and Sample Student Prompts, https://drive.google.com/file/d/1-_wiX1jKxGcO48OgOVM1Zdvy9ZT1Kq-P/view?usp=drive_link

Smartwatch Executive Function Supports for Students with ID and ASD, by Rachel E Wright et al. addresses UDL guidelines to support executive functioning skills such as task initiation, task completion, time management. Three students were part of this study all of whom had intellectual disabilities and two of whom had autism spectrum disorder (Wright, 2022). The students ranged in age from 19-21 years old and were attending a post-secondary education program at a large, public university (Wright, 2022). The goal of the study was to provide increased independence in daily life (Wright, 2022). Before the study, the students had relied on visual support prompting by staff to move from one activity to the nex (Wright, 2022)t. Their dependence on this support made fading difficult (Wright, 2022). The study asked two questions:

Research Question 1: What are the effects of using a wear-able device as a self-operated executive function support option for prompting on independent task completion by college students with ID and ASD?

Research Question 2: Do college students with ID and ASD report their experience using the wearable device as an executive function support system to be beneficial and socially acceptable?

(Wright, 2022, p. 65)

This study used Samsung smartwatches that were able to use the GuruWear application and the MoveUp! Alarm application (Wright, 2022). Students were consulted regarding how they felt about the look and feel of the visuals and trained how to input their appointments themselves (Wright, 2022). Each week, students input three appointments with four steps each (Wright, 2022). The order of the appointment name, date & time, and list of the tasks was kept consistent in its formatting (Wright, 2022). A baseline was established which determined that none of the three students were able to complete their appointed tasks independently (Wright, 2022). With the smartwatch, all three of them achieved a 100% completion rate after 8 sessions (Wright, 2022). The watch was then removed and all three students struggled to maintain any independent task completion (Wright, 2022). When the watch was reintroduced they all returned to 100% task completion (Wright, 2022). The authors concluded that the smartwatch is an effective component of UDL and supports more independent executive functioning (Wright, 2022). Perhaps not surprisingly, they also found that the students felt the devices were socially acceptable (Wright, 2022).

The first image below illustrates checklist directions, student set time schedules, and directions within a schedule (Wright, 2022, p.67). The second image shows the baseline performance of the three students, their success rate with the watch, what happened when the watch was removed and finally the result of reimplementation (Wright, 2022, p.70).

Figure 1. Example screenshots of initial and final formula steps as displayed on the smartwatch platform are the top two images. Lower two images are screenshots of the MoveUp! alarm application alarm configuration view on the smartwatch. Source: Author., https://doi.org/10.1177/0162643420950027

Figure 2. Percentage of appointment tasks completed independently across students with and without the smartwatch intervention., https://doi.org/10.1177/0162643420950027

In another article, Open University in the United Kingdom identified several obstacles to success for their students with a variety of disabilities (Lister, 2021). One substantial barrier that was identified was the actual process for declaring a student’s disability with the college (Lister, 2021). The nature of the forms that needed to be completed fell into executive functioning, visual impairment and other disability impacted areas (Lister, 2021). The author noted that, “Participation is a key principle in disability studies, often summarized as ;nothing about us without us’” (Lister, 2021, p. 2) With this in mind, Open University included 13 students with disabilities in the development of an AI solution, Taylor, the Disability Disclosure Virtual Assistant: a Case Study of Participatory Research with Disabled Students. Taylor can interact in a voice format, converse about the student’s disability and needs, then summarize that conversation for disclosure forms (Lister, 2021). Taylor can also point students towards other university services, clarify academic timelines, and more (Lister, 2021). Both of these examples use multiple means of communication in order to increase accessibility and independence in a socially acceptable way (Lister, 2021). By creating a virtual assistant for a variety of purposes that all students may utilize, embedded with multimodal communication, the university is incorporating elements of universal design (Lister, 2021).

Below is a sample conversation between Taylor and a student with a disability. In this image, Taylor offers advice, answers questions and then summarizes the conversation (Lister, 2021, p. 10)

Taylor, the Disability Disclosure Assistant, https://doi.org/10.3390/educsci11100587

AI is also being used to individualize learning for students with disabilities through computer monitoring of the response of the learner in order to support engagement. In An Evaluation of an Adaptive Learning System Based on Multimodal Affect Recognition for Learners with Intellectual Disabilities, by Standen et al. artificial intelligence tools were used to monitor students for three specific affective states: engagement, frustration and boredom. The 67 students all had intellectual disabilities with and without Autism Spectrum Disorder ranging in age from six to eighteen (Standen et al., 2020) . The students attended school in the United Kingdom, Italy and Spain (Standen et al., 2020) . They were also recommended by their teachers as being able to benefit from the MaTHiSiS system (Standen et al., 2020) . The learning system used was able to determine the students' affective state and adjust both the difficulty level of content presented as well as the learning content in order to maximize the affective state for learning (Standen et al., 2020) . The ideal affective state for learning achievement was determined to be either mild frustration or engagement (Standen et al., 2020). Boredom was associated with lower achievement (Standen et al., 2020). Interestingly, students were able to act as their own control (Standen et al., 2020) . The system alternated between sessions that combined learning achievement with affective state or with learning achievement alone (Standen et al., 2020) . As affective states can sometimes manifest differently in students with disabilities, the students’ teachers recorded videos of them using a variety of different devices and labeled their affective states (Standen et al., 2020) . This video was then used to train machine learning algorithms (Standen et al., 2020) . As illustrated below, the system combined facial expressions, eye gaze, body pose, voice input, gestures and interaction with the learning materials as well as the level of difficulty (Standen et al., 2020, p. 1752).

MaTHiSiS Adaptive Learning Ecosystem, https://doi.org/10.1111/bjet.13010

This study found that they were able to impact engagement, but not achievement (Standen et al., 2020). Their findings aligned with UDL principles in that engagement increased when the learning mode, difficulty level and content were aligned with the individual students (Standen et al., 2020) . The authors recognized that critics would rightly inquire whether increased engagement is success in the absence of increased engagement (Standen et al., 2020) . The researchers recommend longer studies to see if achievement could be positively correlated (Standen et al., 2020).

When considering the risks of AI use for students with or without disabilities, critics often give voice to the concerns covered in, Artificial intelligence: A help or threat to contemporary education. Should students be forced to think and do their tasks independently?, by Xiaohu Xie and Tao Wang, 2023. The authors set out to determine whether or not consistent use of video game learning, virtual and augmented environments, simulations and AI would negatively impact students cognitive abilities and working memory (Wang & Xie, 2023). At Tianshu Normal University in China, 539 students were part of this case study (Wang & Xie, 2023). Of these participants, 322 students were in a music program which heavily incorporated a wide variety of technology augmented learning environments including both AI and video game learning (Wang & Xie, 2023). For the control, 217 students studying visual arts were chosen from a program which did not include these technologies (Wang & Xie, 2023). All participants were given the Weschler Adult Intelligence Scale as the method of measuring any loss of cognitive abilities (Wang & Xie, 2023). The authors address critics of prolonged student use of AI, stating that, “The findings of this study indicate that the use of artificial intelligence (AI) in education, specifically in game learning environments, does not deteriorate the cognitive abilities and memory of students,” (Xie & Wang, 2023, p. 3106). The authors observe that this result will likely heat up the discussion around risks of student dependence on AI as it is difficult to become comfortable with the implementation of such significant change to education and the role of teachers so quickly (Wang & Xie, 2023). They also predict that teachers may gradually shift to the role of facilitators rather than directors of their students’ learning as AI becomes increasingly prevalent in education (Wang & Xie, 2023).

Special education teacher responsibilities are also being impacted by the availability of AI tools (Klein, 2024). The pros and cons of AI in special education, by Alyson Klein, highlights how tasks such as the development of and adherence to students' individualized educational plans can now be supported with Magic School. This is an AI which is specifically designed to support education needs, but teachers should be extremely careful not to include any information that would impact student privacy (Klein, 2024). It is recommended that teachers are confident in both their knowledge of special education and AI use before using the tool as an aid in IEP development (Klein, 2024). This article also addresses critics’ concerns regarding student use of AI,saying that, “Since it’s likely that students will be able to use those tools in the professional world, it makes sense that they begin using them in school,” (Klein, 2024, p. 16)

Privacy is a repeated concern in the article, Elicitation and Empathy with AI-enhanced Adaptive Assistive Technologies (AATs): Towards Sustainable Inclusive Design Method Education, by McDonald et al.. The authors draw attention to the importance of privacy for vulnerable populations (McDonald, 2023). Reflecting on a study with Information Systems graduate students involving the AI applications of Grammarly and Pinata, the researchers found that the future software developers were well aware of their own privacy concerns, but did not have empathy for users with disabilities (McDonald, 2023). Instead, there seemed to be an underlying, erroneous assumption that the benefits of the technology outweighed any risk to the users (McDonald, 2023). This study highlights a larger need for individuals with disabilities to have a voice in both the development of technology as well regulations around use of data and user privacy (McDonald, 2023). Somehow participatory design must be achieved without burdening the communities being served (McDonald, 2023).

Despite these risks, the practical applications to support executive functioning skills in students with disabilities are tremendously exciting. As AI makes it increasingly simpler for schools to embed technologies which makes learning more automatically accessible, theories like Universal Design for Learning can be more consistently and effectively applied. Behaviorism ideals of immediate, consistent feedback become scalable. Wearables like smartwatches create real independence where students can be supported directly by the technology without relying on support staff. As ChatGPT is able to assist students with accessibility, planning written expression, locating resources, practicing conversations, suggesting time management options and much more; executive functioning is also supported by eliminating the need for multiple tools. The opportunity for computer supported learning to individualize instruction is just beginning to be realized and technology where the user's affect is being measured in real time is intriguing. The benefits may outweigh the risks, but the risks should still be mitigated as we venture into this exciting, accessible future.