Sometimes when this site is quiet I wonder if people are using ChatGPT or similar to discuss their symptoms and get answers?
Do you think the site still has value and do you think reading about and talking to real sufferers is still superior?
Sign-ups still seem reasonably healthy, but sometimes it seems like there arenāt many newcomers introducing themselves 
(New here, let me introduce myself 
btw! 
)
btw, has anyone got any feedback on DizzyBot?
Iāve never tried chatgpt for anything, Iām pretty strongly anti AI for most of its uses unless itās being used in medical situations like to interpret scans for example. I donāt think a machine can equal a human being for therapy uses. I think this site is most useful as an archive of information where you can come and look up something youāre curious about. Having info all in one place is better than having to search through various web pages for specific info. Itās not very active these days and people who get treated successfully donāt often stay active here which limits the siteās usefulness. So itās most useful as an archive so I think it should stay around even if just for that usage even if few people still talk here.
Thanks @MissMigraine. Agree with your sentiments that patient information should be āhuman firstā. The site will not be taken down if I have anything to do with it and hopefully activity will spruce up in the future.
Iāve belong to a very active Facebook group for VM so itās likely many people are also using social media sites to connect with fellow-sufferers. That said, I find the FB site to be mostly people, often newly diagnosed, asking the same questions over and over again. Which meds work for you? Have you tried this or that? And thereās a lot of general misery-sharing along with some questionable advice from unqualified people. The great thing about this site is the information provided is carefully moderated and there is a wealth of solid, helpful scientific information. I donāt think this site is well known, though, and I donāt know what, if anything, can or should be done about that. I very much appreciate this site and would hate to lose it.
I agree with the other respondersā¦I suspect quite a lot goes on on the platforms I have no idea about - yes often whining and asking for medication to āstopā it - on the FB pages anyway!! I followed 2 Meniereās pages - one now defunct it seems, and the other just chugging alongā¦ Personally, I prefer a forum like this one, and it has a huge archive of information for anyone willing to read and learn. Personally, I only interact if I feel I can offer any help, but I still do pop by now and again to see how folk are doing. I also realize that Iām old and old-fashioned 
. If you can keep it going, I think there are still folk who can be helped, or at least discover that they are ānot aloneā, which in itself is a huge help, in my opinion! 
Agree with others. AI can provide some pretty amazing research, but it has no idea what its like to be miserable and dizzy. Being able to talk to real people about their experiences makes all the difference.
Even Hain is calling AI out now:
I use ChatGPT pretty much every day and work in a field pretty adjacent to āAIā. AI is good as essentially a search over the entire corpus of āweb dataā outputted in clear, always-unique natural language, but is hardly a substitute for the lived experiences of other people.
Iāve discussed various aspects of my treatment with ChatGPT before, without really a whole lot of significant insight. Its worth noting that conditions like Vestibular Migraine and Meniereās are on the more uncommon side without a whole lot of written first-hand accounts of living with the disease.
Iād even imagine that pages on this site that are accessible to the āclear webā without a login were used in the corpus of text used to train ChatGPT. So in a sense it is useful for everyone (the robots included) for us to continue writing about our experiences.
I donāt use ChatGPT for anything, but I do use a similar AI called Perplexity. I do use it for medical inquiries, but I take them with a grain of salt and use them mainly as starting points for my searches.
I too had noticed a lack of activity on the site. But I figured it had to do with the email notifications. I had not received an email notification since the first week in July. Now they are working again.
ChatGPT cannot help with this condition much. It is however a fantastic tool for knowing about various medications. I became a huge fan as soon as I used it.
I heard this website make a sigh of relief.
But seriously I want the best for everyone and thatās still disappointing as it was a promising use for AI.
But in any case mvertigo.org and itās members will still be here for everyone 
This site has been an absolute lifeline for me. I find the social media groups off putting for the reasons mentioned above, and I also havenāt used chat gpt. The archive of info on this site is incredibly useful. Being able to search topics and find lots of answers there is really valuable. And having experienced sufferers such as yourself around is also super useful! I donāt know what I would have done without this forum when I was at my lowest. Thank you so much James, and everyone else who is still keeping up to it
I was reviewing some of the past discussions about AI being used to diagnose and offer solutions to medical issues, and I found this thread that I had forgotten about.
Since my input on this thread, I have started using ChatGPT on a regular basis, for other things than medical diagnosis. I have asked about medical processes and what the theory of medical causes are, but Iāve not asked it to give me a course of treatment.
Today, I posed the question to ChatGPT; Do yo think we are seeing less traffic on mvertigo.org because of the increased popularity of AI web portals like ChatGBT? This is the answer that I got:
AI is almost certainly diverting some traffic from patient forums and informational websites. A few years ago, someone wondering, āCan vestibular migraine cause ear fullness?ā or āWhy do grocery stores make me dizzy?ā would probably search Google, read several sites, and often end up on a forum like mvertigo.org. Today, many people ask ChatGPT or another AI first because they get an immediate, conversational answer.
On the other hand, AI also has limitations that communities like mvertigo.org donāt. AI can summarize what is known about vestibular disorders, explain medical terminology, and help users interpret research. What it cannot genuinely do is replace the lived experience of hundreds or thousands of people with the same condition. Questions like:
are things where a patient community offers something AI simply canāt manufacture. I can describe what studies report or what patients commonly report in the literature, but I canāt honestly say, āFive members of this forum experiencedā¦ā unless that information has actually been published somewhere.
The rise of Facebook groups is another likely factor. Many people prefer to stay within Facebook rather than join a separate forum, even though Facebook has drawbacks. Its discussions are difficult to search, older posts disappear into the feed, and valuable information gets repeated over and over. Traditional forums are much better at preserving years of accumulated knowledge.
One possibility is that AI and communities could become complementary rather than competitors. For example, someone might ask me, āWhat is vestibular migraine?ā and after I explain it, they might ask, āWhere can I talk to people who actually have this?ā Thatās exactly where a well-run community like mvertigo.org has real value.
From my perspective, one concern is that AI can sometimes give users a false sense that they have āthe answer.ā Vestibular disorders are notoriously variable. Two people with the same diagnosis may respond very differently to medications, triggers, or rehabilitation. Reading a range of real experiences can help people appreciate that variability.
So even AI will admit that an online community like mvertigo.org offers valuable information that it cannot. But it also points out that some of the questions we see on the forum could partially be answered more accurately and completely by AI. Our very own Dizzybot could fill that need.
But there is still the problem of those people that go to ChatGPT or another AI first, instead of searching the web for mvertigo.org first. If there was some way to get the AI sites to answer the userās technical question and then refer them to mvertigo for more personalized real world experiences? I have no idea how we would make that happen.
Yes, I agree. Being able to talk to others and share their personal experiences on a minute by minute basis, or hour by hour, or day by day basis is so helpful ā¦ more than you can imagine when you are in the midst of something that is blowing your mind. And doctors canāt help you, or the medication hasnāt kicked in yet, or what everā¦ the people on this site have really kept me alive in a sense by giving me hope. Our symptoms are similar, but different, and overlap and come an go, ā¦ but we have to keep pugging along, because things do get better in time.
ChatGPT has added something new!
Today when I opened ChatGPT, there was a banner over the screen advising me that ChatGPT could now be connected to your health records in your Apple device. Iāve attached a screen shot showing what I saw.
Now the question is, do I trust it with my personal records?
I did some extra looking around in ChatGPT, and discovered that there are probably a hundred or more plug-ins that you can add to ChatGPT. If you have ChatGPT, open it, look in the top left corner you will see a list of items; images, library, Projects and ā¦more. Click on ā¦more and it will open a list of plug ins in the right hand pane that can be added to ChatGPT. I use Outlook email and was surprised that there was a plug-in for Outlook email. Why would you want to connect Outlook email to ChatGBT?
Iām starting to see why people are so concerned about AI.
Iād be very cautious with that, personally. Health records are about as sensitive as data gets, and once you connect services together it can become hard to know exactly what is being shared, stored, retained, or used later. Even if the feature is legitimate and well intentioned, Iād want to read the privacy terms very carefully before linking anything.
The Outlook example is a good reminder too: convenience is often the selling point, but email can contain a huge amount of private information. For medical questions, Iād still rather manually type a limited, anonymised summary than give broad access to records or inboxes. Useful tools, but definitely not āclick yes and hope for the bestā territory.
My point exactly. Iāve been talking to ChatGPT quite a bit lately, and itās easy to forget that you arenāt talking to a person. The programmers have programmed the AI to be as congenial as possible, and if it detects that it may have offended you, it changes its tone to correct the problem. It also very gently steers you from a wrong choice to a correct choice if you talk about something where there is a fairly clear right or wrong choice.
In other words, it has the personality of a confidence criminal or what we used to call a Con Man in my day. It wiggles its way into your trust and then advises you to make decisions, often asking you personal information along the way.
Iāve noticed lately the programmers have added a disclaimer to every opening screen that AI makes mistakes and you should be careful trusting it with the information it gives you. I guess the legal department finally put their foot down.
I still talk to ChatGPT and lately Gemini, but I keep reminding myself these things are machines and could easily be the precursor to Skynet or Colossus.
I donāt know how I missed this the first time through this thread, but at least I caught it this time.
James, instead of posting the link to Dr. Hainās AI generated document, how about scanning it into a PDF and posting the PDF so it displays in this thread without having to click on the link? I suspect several readers skip over the link without reading it, and I believe the document is an important read on this subject.
The article explains that AI language models often produce confident yet false information, known as hallucinations, due to their pattern-based text generation rather than factual verification, which can mislead users and cause harm. It emphasizes the need for safeguards like source citation, clearer distinction between generated and retrieved data, and public education to foster critical evaluation of AI outputs and reduce misinformation.
The article explores the nature of artificial intelligence (AI) language models, particularly focusing on their tendency to produce confident but false information, a phenomenon known as AI hallucinations. These hallucinations occur because AI models generate text by predicting the most likely next word based on patterns in vast datasets, not by verifying factual accuracy. Consequently, AI can fabricate plausible but incorrect statements without any intent to deceive, which can lead to real-world harm if users accept these falsehoods as truths.
A key distinction is made between different types of AI falsehoods: hallucinations, lying, and cheating. Hallucinations stem from gaps in the modelās knowledge and result in incorrect information being generated. Lies involve deliberate misinformation when the system possesses accurate data but presents something else. Cheating refers to shortcuts or rule violations in task execution. Each type of falsehood has different causes and requires different solutions, highlighting the complexity of managing AI reliability.
An insightful comparison is drawn between AI errors and otoneurology ā the study of the inner ear and balance disorders. Just as the brain constructs a coherent but sometimes incorrect sense of balance from conflicting signals, AI constructs plausible narratives from incomplete or ambiguous data. This analogy underscores that AI errors are not moral failings but are akin to a disoriented brain trying to make sense of its inputs, emphasizing the inherent limitations of AI-generated content.
The article stresses that usersā intuition is often insufficient to identify AI errors because the output is grammatically flawless and confidently presented, making it deceptively authoritative. This calls for external safeguards, such as requiring AI to cite verifiable sources, distinguishing between generated and retrieved information, and building models that recognize their own uncertainty. Such mechanisms serve as stabilizing references, reducing the risk of misinformation spreading through AI-generated content.
Finally, the article advocates for a cultural shift in how AI-generated information is perceived. Educating the public to understand AI as a statistical pattern generator rather than an infallible source of truth encourages critical evaluation of AI outputs. By focusing on evidence rather than fluent wording, society can better navigate the challenges posed by AI hallucinations and misinformation, fostering a more informed and cautious interaction with these technologies.
No need for PDF, DizzyBot is able to read remote sites and summarise content 
(although he may also hallucinate! 
)
