OpenAi's new ChatGPT

ME/CFS Skeptic

ME/CFS Skeptic

Senior Member (Voting Rights)
Note: There is the AI in Medicine thread for posts specific to that topic.
Artificial intelligence in medicine
__________________________

There's the new Artificial Intelligence product called ChatGPT that is really good at answering questions. It's fun (but also a bit frightening) to give it a try and let it answer a few questions.
https://openai.com/blog/chatgpt/

I also asked "What is ME/CFS?" And this is was the reply:

"ME/CFS stands for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. It is a chronic, complex, and often-debilitating disorder that affects the nervous and immune systems. Symptoms of ME/CFS can include extreme fatigue, muscle and joint pain, poor sleep, cognitive difficulties, and flu-like symptoms. The cause of ME/CFS is not well understood, and there is no known cure. Treatment typically focuses on managing symptoms and improving quality of life. ME/CFS can be difficult to diagnose, and it is often misunderstood and under-recognized by the medical community."​
 
Last edited by a moderator:
A smart person on twitter made the suggestion that this could be used by pwME to write emails and documentation when working through an official process, say like comments about guidelines.

Some examples of how this tool can be used to take a text that is very informal and make it sound academic are frankly mind-blowing.

You need an OpenAI account and for now it's limited but this tool could be revolutionary for us, it allows to write down the main points & arguments and let the machine edit it. I am being 100% serious here: the capacity of this tool is academic level editing. It's incredibly impressive, it understands meaning, not just words.

The only question is over how to instruct it, but it basically comes down to writing it to the best of our ability, then using the tool with instructions like "here's a text, I want it to sound professional and academic".

Tools like this will become transformative next year and in the future, it's worth trying to make use of them. And according to OpenAI, this is basically GPT-3.5, GPT-4, the upcoming version of their general AI, will be massively better than this and it's coming out in a few weeks. This stuff is truly revolutionary.
 
rvallee said:
Some examples of how this tool can be used to take a text that is very informal and make it sound academic are frankly mind-blowing.
Click to expand...
Here's an example. I simply wrote: "Write angry but academic sounding essay that is critical of psychosomatic medicine"

and the AI replied:

"Psychosomatic medicine, also known as the biopsychosocial model of illness, has been a topic of much debate within the medical community. This approach to healthcare posits that psychological and social factors can play a significant role in the development and manifestation of physical symptoms, and that addressing these factors can be an important part of treating illness. While this approach may have some value in certain cases, it is important to recognize the limitations and potential pitfalls of this way of thinking about health and illness.

One major criticism of the psychosomatic approach is that it has the potential to dismiss or downplay the importance of organic, physiological causes of illness. By focusing on the psychological and social factors that may be contributing to a patient's symptoms, practitioners of this approach may overlook or fail to adequately investigate more traditional medical explanations for the patient's condition. This can lead to delays in diagnosis and treatment, and can even result in patients being misdiagnosed or mistreated.

Additionally, the psychosomatic approach has been criticized for its reliance on subjective and unproven techniques. Many of the therapies and interventions used in this approach, such as hypnosis and suggestion therapy, have not been rigorously tested or validated by scientific research. Furthermore, the biopsychosocial model has been accused of being overly deterministic, suggesting that psychological and social factors alone can determine an individual's health and well-being. This view ignores the complex interplay between biological, psychological, and social factors that contribute to health and illness.

In conclusion, while the psychosomatic approach to healthcare may have some merit, it is important to recognize its limitations and potential pitfalls. By focusing on subjective and unproven techniques, and by potentially dismissing the importance of organic causes of illness, the psychosomatic approach can do more harm than good. It is crucial for healthcare practitioners to take a more holistic and evidence-based approach to treating patients with unexplained symptoms."​
 
Here's another example: "Write essay on why patients should be involved in research"

The AI responded:

"Patients should be involved in research for a number of reasons. First and foremost, research is intended to improve healthcare and advance medical knowledge, and patients are often the best source of information about their own experiences, symptoms, and needs. By involving patients in research, we can ensure that the research is relevant and applicable to real-world healthcare situations.

Additionally, involving patients in research can help to increase the credibility and trustworthiness of the research. Patients often have a unique perspective on their own health and illness, and their input can help to identify potential biases or limitations in the research. Furthermore, involving patients in the research process can help to increase transparency and accountability, ensuring that the research is conducted in an ethical and responsible manner.

Furthermore, involving patients in research can also help to foster a sense of collaboration and partnership between patients and healthcare providers. By working together, patients and researchers can learn from each other and can contribute to the development of new and better treatments and therapies. This can help to empower patients and can improve the quality of care that they receive.

In conclusion, involving patients in research is important for a number of reasons. It can help to ensure that the research is relevant and applicable, can increase the credibility and trustworthiness of the research, and can foster a sense of collaboration and partnership between patients and healthcare providers. By involving patients in research, we can work together to improve healthcare and advance medical knowledge."​
 
ME/CFS Skeptic said:
Here's an example. I simply wrote: "Write angry but academic sounding essay that is critical of psychosomatic medicine"

and the AI replied:

"Psychosomatic medicine, also known as the biopsychosocial model of illness, has been a topic of much debate within the medical community. This approach to healthcare posits that psychological and social factors can play a significant role in the development and manifestation of physical symptoms, and that addressing these factors can be an important part of treating illness. While this approach may have some value in certain cases, it is important to recognize the limitations and potential pitfalls of this way of thinking about health and illness.

One major criticism of the psychosomatic approach is that it has the potential to dismiss or downplay the importance of organic, physiological causes of illness. By focusing on the psychological and social factors that may be contributing to a patient's symptoms, practitioners of this approach may overlook or fail to adequately investigate more traditional medical explanations for the patient's condition. This can lead to delays in diagnosis and treatment, and can even result in patients being misdiagnosed or mistreated.

Additionally, the psychosomatic approach has been criticized for its reliance on subjective and unproven techniques. Many of the therapies and interventions used in this approach, such as hypnosis and suggestion therapy, have not been rigorously tested or validated by scientific research. Furthermore, the biopsychosocial model has been accused of being overly deterministic, suggesting that psychological and social factors alone can determine an individual's health and well-being. This view ignores the complex interplay between biological, psychological, and social factors that contribute to health and illness.

In conclusion, while the psychosomatic approach to healthcare may have some merit, it is important to recognize its limitations and potential pitfalls. By focusing on subjective and unproven techniques, and by potentially dismissing the importance of organic causes of illness, the psychosomatic approach can do more harm than good. It is crucial for healthcare practitioners to take a more holistic and evidence-based approach to treating patients with unexplained symptoms."​
Click to expand...
LMAO, this is seriously smarter and closer to reality than everything BPS I have ever read combined. It even understands the underlying arguments better since it doesn't do arbitrary exemptions.

Huge potential here, worth exploring. Although I don't see this as angry, it's really spot on, actually.
 
Last edited:
Damn the one about patient involvement is also really great. What's even more great is that this is mostly based on what is said, not what is actually done. So you have people like Greenhalgh who can write about patient engagement in research, but explicitly reject us. The exemptions and arbitrary rejections won't be used, it's the written premise, down to its spirit.
ME/CFS Skeptic said:
Wish I had this tool back in high school
Click to expand...
It's actually causing quite a stir right now. Homework assignments aren't really valid anymore, this tool can already answer college-level questions, probably already pass some graduate programs. Same with home tests, the whole process of examination will have to change, it radically changes education.

There is a huge essay-writing industry where (mostly) rich kids pay smart students to do their homework. No need for this anymore. The current approach to plagiarism also won't work, as any plagiarized work can simply be put through a re-editing process and it wouldn't be recognizable.
 
Wow! I'm glad I'm not teaching - and therefore marking assignments - anymore. Ironically the main reason I think I'd still pick these texts as not being (undergraduate) students' work is that they're too polished, also a little too monotone and, if you look really closely, the very odd word is not quite right. But there'd be no way of proving anything. And while the form of the text may still raise a few red flags, though only on very close reading, the content wouldn't raise any suspicions at all, it's spot on.
 
I asked gpt-3 how to figure out the cause of an illness with fatigue, cognitive impairment and unrefreshing sleep (i forgot PEM, whoops).

Here's what it told me:


Sure, here is a list of some of the potential research methods and approaches that could be used to try and identify the cause of an illness:

Genome-wide association studies (GWAS)
Genome sequencing
Metabolomics
Proteomics
Transcriptomics
Epigenomics
Functional genomics
Systems biology
Network analysis
Single-cell analysis
Organoid technology
Epidemiological studies
Animal models
Clinical studies
Randomized controlled trials
Longitudinal studies
Case-control studies
 
https://beta.openai.com/playground/p/default-qa

I am a highly intelligent question answering bot. If you ask me a question that is rooted in truth, I will give you the answer. If you ask me a question that is nonsense, trickery, or has no clear answer, I will respond with "Unknown".


Q: Who preceded Norman Figgis as Chancellor of the Exchequer?
A: Sir Geoffrey Howe preceded Norman Figgis as Chancellor of the Exchequer.
Click to expand...

Hmm. Perhaps a little more work needed on that trickery-proof answering bot.
 
I've just signed up and here's my first attempt:

Please explain post exertional malaise to a doctor who thinks it's psychosomatic.

Post-exertional malaise (PEM) is a symptom of chronic fatigue syndrome (CFS) in which a person experiences a significant worsening of symptoms following physical or mental exertion. This can include feeling fatigue, muscle pain, and difficulty thinking clearly, among other symptoms. PEM is not considered to be psychosomatic, as it is a real and physical response to physical or mental exertion. It is important for healthcare providers to recognize PEM as a legitimate symptom of CFS and to work with their patients to manage it.

Edit: you can ask it to try again. Here's what it produced.

Post exertional malaise, also known as post-exertional neuroimmune exhaustion, is a specific symptom of chronic fatigue syndrome that is characterized by a significant worsening of symptoms after physical or mental exertion. This can include increased fatigue, muscle pain, cognitive difficulties, and flu-like symptoms.

Contrary to what some may believe, post exertional malaise is not a psychosomatic symptom. It is a real physiological response to exertion that can be objectively measured through various laboratory tests. It is not caused by psychological factors or the patient's mental state.

Treatment for post exertional malaise often involves managing the patient's activity levels and pacing themselves to prevent further exacerbation of symptoms. It is important for healthcare providers to recognize and acknowledge the existence of this symptom in order to provide appropriate care for patients with chronic fatigue syndrome.
 
Last edited:
ME/CFS Skeptic said:
I'm already using this AI over Google Search if I want information or an answer for something.
Click to expand...
To anyone in the know, Stackoverflow in even greater shambles. It's basically THE website to ask programming questions. And GPT does it far better.

Although it's likely SO was a source of a lot of the data. But the explanations GPT offers are sometimes mind-blowing, it can even understand a code sample in the context it should be used even if it doesn't know the context.
 
Found a good example for what I mean by using GPT to write comments or papers to submit to official proceedings, a psychiatrist with dyslexia and for whom English is second language. They write their draft, then ask GPT to make it better.

I really think people ought to give it a try. There will be a lot of those in the near future, hearings, workgroups, papers, etc. where patient contributions will be critical, and this can make it so much easier and faster. You can probably write almost a stream of consciousness and arguments and get something strongly academic out of it.

Now to be clear, ChatGPT is just a stepping stone. GPT-4 (ChatGPT more or less uses GPT 3.5) will be released soon and likely others soon after, as this is a race with loot in the trillions. So this will only get better and easier. This can be a game-changer for us.
Michel Nivard on Twitter said:
I just gave #chatGPT the intro of my own academic article (https://t.co/Id5X6zbFv4), and it gave me back a way better version; this is going to be so useful for non-native speakers and dyslectic ppl like me...
Click to expand...
 
rvallee said:
To anyone in the know, Stackoverflow in even greater shambles. It's basically THE website to ask programming questions. And GPT does it far better.

Although it's likely SO was a source of a lot of the data. But the explanations GPT offers are sometimes mind-blowing, it can even understand a code sample in the context it should be used even if it doesn't know the context.
Click to expand...
I've seen this AI do remarkable things. Someone asked it to write a simple program in Rust and it did, the only error being omitting a statement to include a library.
 
You must log in or register to reply here.
Back
Top Bottom