chatgpt for supporting cfs research and as tool to assist the sickend?

[linusbert]

Neither is available in my country .
But the version for Android us available with a limited number of free questions (the AI is moderately smart). It told me there is no way to cure my post-nasal drip

just use brave browser, it comes with TOR integrated. it also has a paid vpn feature.
if this doesnt work, checkout OPERA browser, it also has a VPN feature.
if that also does not work, buy nordvpn or any other vpn service and just connect. there are browser plugins for vpn so you do not even have to install it.

 

[Hip]

It would be fun to join a forum like say Reddit, and whenever anyone asks a question, simply post the answer that ChatGPT provides to that question!

I wonder if anyone would ever work out that they were dealing with AI!



As an aside, if anyone here has any curious medical symptoms that they would like diagnosed by AI, then Symptoma is one of the most intelligent and accurate online AI diagnostics tools. A study compared all the available AI diagnostic tools, and Symptoma came out as the best.

Symptoma is free and needs no registration or personal ID details; just enter your symptoms, describing them with normal natural language, then answer any questions Symptoma asks you, and it will provide a list of possible diagnoses, with the most likely diagnosis at the top of the list.

If you type in your ME/CFS symptoms, it very quickly diagnoses you with ME/CFS.

 

[BrightCandle]

If you type in your ME/CFS symptoms, it very quickly diagnoses you with ME/CFS.

It is notable that AI/expert systems can provide a diagnosis in minutes with a few exchanged questions and answers and the entire NHS, 109 medical staff and 187 appointments can't. The AI is so incredibly better than the doctors and nurses now its hilarious. Even ChatGPT gives out a brilliant description of Long Covid and can diagnose it and its not even specialised for that purpose, yet these same people have received "anxiety" from their doctors. Its a stunning display of how useless the medical profession has become and its now just gatekeeping access to medicine.

 

[Andryr]

If you type in your ME/CFS symptoms, it very quickly diagnoses you with ME/CFS.

The first time Symptoma diagnosed me with a bunch of psychiatric disorders although it had asked me if I'd had depression, bipolar disorder etc. (my answers were negative). Another "you are nuts" case.
The 2nd time CFS was in on the 5th position of possible causes list. The rest of Top-5 was filled with heart and blood issues although it had explicitly asked me questions about my blood analises and my heart condition.
Verdict: questionable.

 

[Hip]

The first time Symptoma diagnosed me with a bunch of psychiatric disorders although it had asked me if I'd had depression, bipolar disorder etc. (my answers were negative).

That's strange, every time I've tried Symptoma, an ME/CFS diagnosis always came at the top of the list, even when I included symptoms such as depression.

Eg, when I enter: fatigue, brain fog, sleep disturbance, depression, then Symptoma comes up with ME/CFS as the first choice.

What symptoms are you entering?

 

[Andryr]

Now I understand I did it wrong. I entered only fatigue and let it ask me the rest of questions about my simptoms.

 

[Hip]

Now I understand I did it wrong. I entered only fatigue and let it ask me the rest of questions about my simptoms.

You can enter lots of symptoms, just separate each symptom with a comma. The more symptoms you enter, the better the results.

 

[Murph]

ChatGPT is a system that produces natural sounding language, not true facts.

A great way to get it to show you how wrong it is is to ask for transit drections to somewhere you know well. It will tell you to get on a non-existent bus. If you know that bus doesn't exist, you're fine. If you don't know, you might believe it.

You can also ask it for facts about things that don't exist, and it will tell you.

What architectural style is Moondawup train station?

Moondawup train station is of a Victorian Italianate style.

It's the same with science. It will tell you things in language so fluent you'd never consider they're completely wrong.

This scientist got it to do some very deep work , and it was all 100% wrong because the topic doesn't exist!

https://twitter.com/i/web/status/1599893818186543105

 

[linusbert]

today i changed to curie (previous version) from davinci (chatgpt),
asked in german what the reasons for copper deficiency were, she said:

"The reason for the copper shortage can have various causes, e.g. the exploitation of copper in the electronics industry or the deforestation of copper mountains."

 

[vision blue]

@linusbert i got a similar complete category misinterpretation when i asked about the role of the rccx module in cfs

After playing aroud for two hours , Im not impressed and see no use for it. even its poem on vertigo i asked for pissed me off and i didnt understand thr lighbulb joke on the tooics i gave it
thonappreciate the effort.

Overall for science questions i find it slows one down having to unravel the story it weaves together. It will btw provide a reference list if you ask for one.
I also found it not at all fun. Reminds me a bit of the game/assignment where you're given 10 words and you have to write a story that uses all 10 words

 

[linusbert]

i think the public chatgpt is garbage for medical questions. can you give me your exact question , i paste it into my chatgpt and send you result. i usually do get good answers.

I also found it not at all fun. Reminds me a bit of the game/assignment where you're given 10 words and you have to write a story that uses all 10 words

you must compare the alternatives, ask a doctor who gives you nonsense answer as well, even worse, and even tries to put you into psychiatry?

 

[lint7]

ChatGPT is a system that produces natural sounding language, not true facts.

A great way to get it to show you how wrong it is is to ask for transit drections to somewhere you know well. It will tell you to get on a non-existent bus. If you know that bus doesn't exist, you're fine. If you don't know, you might believe it.

You can also ask it for facts about things that don't exist, and it will tell you.

What architectural style is Moondawup train station?

Moondawup train station is of a Victorian Italianate style.

It's the same with science. It will tell you things in language so fluent you'd never consider they're completely wrong.

This scientist got it to do some very deep work , and it was all 100% wrong because the topic doesn't exist!

https://twitter.com/i/web/status/1599893818186543105

In a lot of cases it is basically a nonsense generator. If you ask it to cite sources it will just present you will papers and authors that don't actually exist.

 

[linusbert]

In a lot of cases it is basically a nonsense generator. If you ask it to cite sources it will just present you will papers and authors that don't actually exist.

sounds like a real human too me ^^

 

[Shanti1]

In a lot of cases it is basically a nonsense generator. If you ask it to cite sources it will just present you will papers and authors that don't actually exist.

Is that why I could actually find the papers used in the references ChatGPT gave me? Wow that is disappointing. Do you know if it will give valid references if they exist? Is there any way to avoid the bunk references? Does this happen in the developer playground as well? (sorry so many questions, this is all new to me!)

 

[linusbert]

Is that why I could actually find the papers used in the references ChatGPT gave me? Wow that is disappointing. Do you know if it will give valid references if they exist? Is there any way to avoid the bunk references? Does this happen in the developer playground as well? (sorry so many questions, this is all new to me!)

happens in dev playground as well.

i think its just how it operates for now. its not precise.

 

[vision blue]

Is that why I could actually find the papers used in the references ChatGPT gave me? Wow that is disappointing. Do you know if it will give valid references if they exist? Is there any way to avoid the bunk references? Does this happen in the developer playground as well? (sorry so many questions, this is all new to me!)

I dont k ow why the poster said it made up references. That certainly was not my experience. In the request i said. Inude all references for inclusion in a scientific journal article. It provided the references and i went to
A few of the papers of most interest. References also were from the appropriate academic literature tho did miss my fave reference on the topuc.

Thats not a defense of chat gpt. I do not find it does a good enough job of distilling the true point of scientific findings. I find even wikipedia to be better. .But fir references, that wasnt the problem.

 

[vision blue]

In a lot of cases it is basically a nonsense generator. If you ask it to cite sources it will just present you will papers and authors that don't actually exist.

Do you have an example of a made up list of authors and papers? I did not find that. Maybe something in the request allowed for that?

 

[Shanti1]

I dont k ow why the poster said it made up references. Tgat certainly was not my experience. In the request i said. Inude all references for inclusion in a scientific journal article. It provided the references and i went to
A few of the papers of most i terest

Thats not a defense of chat gpt. I do not find it does a good enough job of distilling the true point of scientific findings. I find even wikipedia to be better. .

I'm not sure under which circumstances it posts made-up references and when it doesn't, but I'm starting to think it is when you ask it for references for which there aren't a lot in existence. For example, today I got references that did not exist when I asked:

"How do carbonic anhydrase inhibitors help with tinnitus. Provide references."

I also got fake references when I asked:

"What antioxidants can be helpful for 8-deoxy-2'hydroxyguanosine? provide references."
I got 7 references, none of which have a match in pubmed or google They seem to be a hodgepodge of partial actual journal article names. Only the last reference actually matches a real publication name, but the journal and author do match the real publication.

Here are the references I was given:

Here are a few references:

1. Vitamin C:
   • Song, Y. J., Lee, J. H., & Kim, J. H. (2015). Vitamin C supplementation decreases oxidative stress and improves antioxidant status in patients with major depression: a randomized, double-blind, placebo-controlled trial. Nutritional neuroscience, 18(7), 269-274.
2. Vitamin E:
   • Kim, Y. J., Park, E. J., Kim, J. H., Kim, K. W., & Kim, H. T. (2010). Vitamin E supplementation reduces oxidative stress and 8-hydroxy-2'-deoxyguanosine levels in elderly people. Nutrition, 26(2), 174-179.
3. N-acetyl cysteine (NAC):
   • Peruzza, S., Cavallini, G. C., Rossi, R., Bonanni, A., Rizzo, M., & Montorsi, F. (2007). N-acetyl-cysteine supplementation reduces oxidative stress and improves sperm quality in infertile men. Urologia internationalis, 78(1), 62-66.
4. Melatonin:
   • Sharaf, M., Farid, K., & El-Gendy, O. (2017). Protective effect of melatonin against oxidative stress and oxidative DNA damage in chronic exposure to lead acetate. Journal of clinical and experimental medicine, 9(2), 33.
5. Coenzyme Q10 (CoQ10):
   • Vafeiadou, K., Altowaya, S., Boucher, J., & Williams, C. M. (2010). Coenzyme Q10 supplementation improves metabolic parameters in obese men: a pilot study. Nutrition & metabolism, 7(1), 7.
6. Glutathione:
   • Mocchegiani, E., & Giacconi, R. (2001). Glutathione and the aging process. Experimental gerontology, 36(5), 727-744.
7. Lipoic acid:
   • Zempleni, J., & Hassan, Y. I. (2017). Lipoic acid as a dietary supplement: molecular mechanisms and therapeutic potential. Oxidative medicine and cellular longevity, 2017.

 

[vision blue]

I'm not sure under which circumstances it posts made-up references and when it doesn't, but I'm starting to think it is when you ask it for references for which there aren't a lot in existence. For example, today I got references that did not exist when I asked:

"How do carbonic anhydrase inhibitors help with tinnitus. Provide references."

I also got fake references when I asked:

"What antioxidants can be helpful for 8-deoxy-2'hydroxyguanosine? provide references."
I got 7 references, none of which have a match in pubmed or google They seem to be a hodgepodge of partial actual journal article names. Only the last reference actually matches a real publication name, but the journal and author do match the real publication.

Here are the references I was given:

I. wonder why we are both researching OGG1!

Will check my list of refs again. Maybe i just got lucky on the ones i went to?

 

[Shanti1]

I. wonder why we are both researching OGG1!

Haha, that's funny. I think of it as a rather obscure marker. I was researching it for a work project involving its measurement in urine. I did have mine tested though, was normal.

Will check my list of refs again. Maybe i just got lucky on the ones i went to?

Let us know!