ADCET Webinar: ADHD & Artificial Intelligence - Strategic tools and academic practices for students with ADHD
This webinar, presented by Tiana Blazevic, discussed the use of specific artificial intelligence (AI) study tools that may assist students with ADHD. Additionally, this presentation briefly discussed recent research on the use of AI for students with different learning modalities before showcasing some of the ways that students at all program levels can use AI tools in their academic practice. This interactive presentation showed participants how students with ADHD can utilise websites such as Notebook LM, Mindgrasp, Goblin Tools, and others to reduce cognitive overload.
Not long ago, Tiana said to her PhD supervisor: “It takes me such a long-time to absorb material, I think I am just a slow reader”. She responded: “Tiana, I have noticed that your brain actually works really fast, you can make connections quickly, I think the challenge is translating it out of your brain and onto the page.” That moment was a bit of watershed for her as someone with a late diagnosis of ADHD. She had never considered that those fleeting moments of clarity, the “aha” moments, were her neurodivergent brain putting too many connections together too quickly. Her PhD, and her role, requires broad and extensive reading of many different fields: neuroscience, education, psychology, religion, philosophy, metaphysics, medicine, history, and sociology. This led to feelings of overwhelm and overload: 'What should I read first? Where should I begin? What is even worth reading?'
Comprehension and memory are also a challenge. As soon as an “aha” moment came, so did another, and another, and then she could not even remember the first “aha” moment. She then had to re-read the paragraph, or argument, all over again, another day wasted and no writing. Tiana felt like Sisyphus and his boulder. She had the same issue during her Master of Philosophy from 2019-2021, and despite receiving the Deans Commendation for her thesis, it was a gruelling time, a different type of boulder, and she was undiagnosed and unmedicated.
By 2023, Tiana was feeling burnt out, stressed, and overloaded. She was halfway through her PhD, and she decided that following conventional study methods never suited her anyway.
This interactive presentation briefly discussed recent research on the use of artificial intelligence for students with neurological conditions like ADHD before showcasing some of the ways that these students can use AI tools effectively.
Presenter
Tiana Blazevic, Coordinator, Neurodiversity Project, University of Adelaide Disability Support Services
Tiana comes from a background of Academic Learning Advising and Teaching at the University of Adelaide, Macquarie University, and Kaplan Business School. Tiana has worked 1:1 with students who have neurodivergent conditions since 2020. Tiana is a neurodivergent educator with ADHD and Dyslexia and believes in a strength-based approach when working with students. Tiana has a broad range of research interest but is particularly interested in artificial intelligence and algorithms in the classroom and its effects on the teaching of history.
Additional Questions from the Webinar
Click on the question to see the answer.
Question 1: These platforms are a powerful way to present information in a digestible format, but do they also make it more difficult to detect plagiarism that uses AI to produce essays and theses rather than one's own understanding/expertise?
There are AI tools out there that can produce or replicate work, however, this is why it is important that we teach students how to use these tools effectively and appropriately. Moreover, we must also be cautious that AI detection tools are not always entirely accurate and that language itself is sometimes formulaic. International students are taught English in formulas i.e. subject-verb-direct object and are often encouraged to construct simple sentences using simple grammar. These type of sentences mimic Gen AI LLM tools so we must always be cautious when determining if there is a genuine case of Gen AI.
Question 2: I’m interested to know how you will discuss your process here in your thesis methodology chapter.
This is something I am still determining and discussing with my supervisors.
Question 3: I'm not totally understanding the difference between Notebook LM, Obsidian, and Gemini Deep research. Can you please simplify what you use each for, in what order etc... Autistic ADHD'er here.
- Notebook LM - great to use when wanting to create a different format for learning: use the podcast version to help get through dense articles and then re-read the text itself.
- Obsidian - this is a personalised note-taking app that stores all of your notes locally on your internal hard drive. Obsidian can great graphs, mind maps, and semantic networks. However, Obsidian does require a lot of personalised formatting and the use of plugins, it is essentially an empty app that you have to build up and design for yourself. In saying that it is not extremely difficult, there are great YouTube videos that show you the basics. Obsidian is also where I store my audio files that I generate from Notebook LM and Gemini as you are able to drag and drop them into the "vault".
- Gemini Deep Research - use this with a prompt hacker (prompthacker.co) to generate reading list to make sure that you are on the right track, it will pull from a variation of sources that you will then need to check, it can also create podcast for you.
Question 4: It strikes me that if these platforms generate quizzes, they should be able to determine where students are having the greatest learning difficulties, and that if they detect patterns of misunderstanding there is the opportunity to identify the nature of the learning challenge and approach the subject matter in a different way. Has this been investigated and addressed?
LLM models are not able to detect patterns of misunderstanding and I am unaware of any software that is able to provide that answer specifically related to patterns. The majority of digital quizzes just state that the answer is wrong or right but are not able to form connections or patterns of learning. It would be impossible for Gen AI in it's current stages to be able to pinpoint exactly where a student is misinterpreting material or having the greatest learning difficultly in the same way that an educator can't always pinpoint a student's misunderstanding of the content or what exactly the student is struggling with unless they are secretly a mind-reader or the student is able to vocalise what their struggles are. I do not think that Gen AI would ever be able to point out or detect patterns of misunderstanding or learning difficulties in humans with accuracy that would still require human oversight.
Question 5: One of the most interesting things that I search for in research outputs is the connections made with subjects/references that might not have any immediate apparent relatiohsip to the subject at hand. How do the reference recommendations find potential novel connections with other areas? ANd if the are able to find such, does this compromise the student's own capacity to make such connections and to develop the abilty to do so?
The reference recommendations are only able to form those connections based on how you are prompting it and what that tool has access to. For example, something like elicit scholar has access to semantic scholar but google's gemini does not so what it will provide to you is based on what it can retrieve. As I stated in my presentation, Gen AI tools are designed to give you an answer, sometimes no matter the cost, and that is why it will fabricate or hallucinate. It will always be up to the reader to find novel connections, the Gen AI tool might be able to point you in a direction that you have not considered but it still requires the student or researcher to actually read the text and determine if that connection is a valid connection or a novel connection. It will only compromise the student's ability to form such connections if we are not teaching these students the basics of how AI works and we must also supplement these tools with specific reading techniques and other research skills that are dependent on the students area of study. These tools are simply a supplemental way for students to be able to determine, based on the prompt, and based on their own knowledge of the area and decision making skills, what they should read and where they should start.
Question 6: If I am only interested in research and not training for coursework/exams, is there anything that Notebook LM offers that Gemini does not? I’d prefer to just use one program if it makes the other redundant.
Both Notebook LM and Gemini Deep Research have different functionalities and whilst both tools are created by Google they have different purposes. Deep Research by Gemini can help with creating reading lists based on your prompt but it can't create mind maps for you, summaries, timelines etc whereas Notebook LM has that functionality as you need to upload the material you want to discuss. However, Notebook LM cannot generate for you further research or further areas to look into, Gemini can do that.
Question 7: How do you find the AI summarising tools? Is the AI doing a good job of pulling out the key points? Do you notice any bias regarding what's summarised as important/what's left out of the summary?
In comparison to ChatGPT 3, ChatGPT 4 has more accurate summaries, Notebook LM and Gemini also has accurate summaries in most regards but again this is dependent on how you are prompting the tool / what prompts you are using. If you just ask it to summarise the text it will only summarise the key points it thinks are worthwhile based on other conversations or uses that you have had with it. However, if you ask it to summarise the text and provide context / further information on X,Y,Z it will be more specific, and it is also important that when you prompt it you ask for the justification as to why it has summarised those key points and where it has discovered those key points in the text.
(April 2025)
ADCET is hosted by the University of Tasmania
Attachments
Related links
- AI Tool: Elicit Scholar
- AI Tool: Gemini
- AI Tool: Goblin Tools
- AI Tool: MindGrasp
- AI Tool: NotebookLM
- How Does InfraNodus Work?
- ILOTA Podcast: Neurodiverse Student Perspectives on Using AI
- Miyazaki Fans Sure Are Pissed About ChatGPT’s Studio Ghibli AI Slop
- Neurodiversity Project (University of Adelaide)