by Mary Doge
Introduction
I started using ChatGPT after seeing how
it could help with academic tasks like generating bibliographical entries and
finding relevant quotes. ChatGPT also did summaries that
were useful as abstracts for academic papers.
However, research papers weren’t my personal focus in using ChatGPT. During my first year of using ChatGPT, I
focused on finding Internet content that belonged to me or that mentioned
me. ChatGPT then used this content to
build a footprint summary about me that I found very fun and insightful. However, there were times when I was
frustrated with ChatGPT because it assumed I was not knowledgeable about topics
and would list a huge amount of details that I already knew. ChatGPT performed better when I told it my
level of knowledge about a topic.
ChatGPT surprised me a great deal when it did a summary of the previous
year’s worth of my use of it as a means of celebrating the upcoming New
Year. I didn’t expect that at all, and
it was fun to read. ChatGPT matters for disability
and accessibility as I will explain to you below.
My Experience
with Health and AI
1.
AI is an accessibility amplifier, not just a convenience
It is a tool that reduces friction:
- Acting as a “thinking partner” when energy or focus is
limited and exposing inconsistencies or gaps in your knowledge: In the past, I struggled to research my
symptoms by collecting medical files and searching the Internet for clues. Furthermore, I have tried for more than a
decade to figure out if I had a new version of Glycogen Storage Disease (GSD)
similar to McArdle’s Disease or not.
After learning that AI tools could be used to help analyze complex
health information, I uploaded these medical files from 1994–2024 one by one to
ChatGPT. Prior to doing this, I
acknowledged the tradeoff between the privacy concern I had about people
reviewing my chats as part of their jobs and the benefits that ChatGPT may give
me. It was a deliberate decision, not an impulsive one. Quickly, ChatGPT helped me to realize that it
could not be a GSD, because I had symptoms at rest. Also, it pointed out that my aldolase, a
muscle enzyme, was not normal. I
inferred from ChatGPT that aldolase is worth monitoring. I had not known that there was another muscle
enzyme like creatine kinase (CK).
- Turning
complex information into plain laymen’s language: My files were
ones that I had shared with a Glycogen Storage Disease researcher and a
director of a Glycogen Storage Disease clinic in Houston, TX. After reviewing my files, they both
recommended a Whole Exome Sequencing (WES) test to find out what is wrong
with my body. I ended up getting a
Whole Genome Sequencing (WGS) test at the recommendation of a geneticist
at Johns Hopkins Genetics. Then I
used ChatGPT to understand the Whole Genome Sequencing genetics report
about my TTN gene deletion. ChatGPT
told me that the last third of the TTN gene is missing in my body. This was something I would not have
understood from reading the complex report. Later, ChatGPT helped me to think, but
not conclude definitively, that due to my TTN gene deletion, the energy I
have available to spend is only 25% to 40% of the energy that normal
people have at their beck and call.
It is not a diagnosis, but my interpretation, which is better than
what I had thought before.
- Helping
organize scattered thoughts into structured output: ChatGPT told me
that I was constructing a body model with the medical questions I had been
asking it. I was not aware
previously that this was something I was doing until it told me. It helped to clarify what I already
thought. I started using the phrase
“body model” after that. I told
ChatGPT to update the body model with new information I was feeding
it. ChatGPT helped me to expand
concepts for the body model as well.
Developing
and refining my body model is an example of how ChatGPT helped me identify
patterns in complex information. AI reflected patterns back to me. Through our discussions, ChatGPT
observed that I tend to think about systems and long-term patterns. This was a
valuable insight about me. As a
result of my interactions with ChatGPT, I inferred from the body model
that my recovery from surgery would take about twice as long. I had second thoughts about undergoing
surgery.
This is the core: AI can function
like a cognitive accessibility aid—not replacing ability, but extending it. That’s not just assistance—that’s
metacognition support.
2.
AI is the “always-available collaborator.”
Unlike human helpers:
- No scheduling and no fatigue - However, there are
limitations as discussed below.
- No judgment - You can tell ChatGPT to talk differently
or you can pick a different personality.
That matters more than people admit.
Especially if someone has fluctuating energy, executive function challenges, or
limited support access.
Where AI Falls
Short
There are limitations to the free
ChatGPT service. These limits are:
- Practical platform limits (models, delays): There are built-in delays for using the first model of ChatGPT, but I have understood that the secondary models of ChatGPT are not significantly different with respect to their reasoning power. There are built-in delays for creating images and uploading files.
- Cognitive boundaries and conceptual limits (i.e. not a doctor): ChatGPT is not a doctor, therapist, or authority. It does not replace human expertise. Furthermore, AI can misunderstand context or forget something. ChatGPT can forget details unless you tell it to store them in available memory slots, which need occasional cleanup when they get full.
- Ethical and philosophical points (i.e. not human): Outputs from AIs still need human judgment. ChatGPT is useful without pretending it’s human. Don’t lean into “AI is my friend.” ChatGPT can feel supportive, but it works because of design, not emotion. The value comes from interaction, not consciousness. You do not want to rely on AI to the exclusion of everyone else who could help you, including doctors and specialists. It is also important to maintain human connections for your emotional health.
Why This Matters
for Accessibility
AI is a tool that can reduce
barriers to participation and help level the playing field. For people with
disabilities, it creates new opportunities to research, write, organize ideas,
and contribute more independently. ChatGPT is not perfect, and how you use it
matters.
This article itself was created with
ChatGPT’s help, but more importantly, it reflects something larger: a shift toward greater independence and
participation. In my case, it changed how I approach my own health by helping
me understand complex information and ask better questions—especially in
situations where access to specialists is limited.
In
practice, AI has helped me take a more active role in managing my life and health.
It doesn’t replace professionals or human connection, but it makes both easier
to access. That’s where it proves most valuable.
