The AI has learned Medicine: Ask it some questions about medical science or psychology

[Hans Blaster]

That is cool. Physics is probably one of my favorite branches of science, and it is cool that you worked in a secret field of physics.

The field isn't secret, I've just chosen not to share.

It does seem that the AI output is not filling or intellectual. Plus, the AI that I use is only 7 billion parameters. If one wants good output, I'd need an AI with at least 10x that and a beefy PC or server rack.

Your AI output isn't any more or less satisfying than others. It's a general problem of the genre. Your limitation seems to be on how much it can digest first.

 

[AlexB23]

The field isn't secret, I've just chosen not to share.

Your AI output isn't any more or less satisfying than others. It's a general problem of the genre. Your limitation seems to be on how much it can digest first.

Yeah, I understand. It is perfectly okay not to share which field you work in, cos we do not want the Russians or Chinese to take your scientific secrets. What the AI could be useful for is to maybe help narrow down what possible diseases one has. I have even installed psychology documents onto the AI.

The AI is limited by a few factors:
1. Date of knowledge: Most textbooks that the AI has learned are from before the 2020s.
2. Limited documents: It would take years for me to find a bunch of PDFs covering all of medical science, so sticking to a few textbooks makes more sense.
3. Limited parameters: The more parameters means the AI has been pre-trained on more data before released to the public. More parameters = "smarter AI"

This is a prompt to the machine to analyze and help a fictional person with her struggles at 3M Company in 2030:

This is what the machine replied with:

 

[timewerx]

The AI that I run is only 7 billion parameters, cos my laptop only has 8 gigs of RAM. If I had 16, 32 or 64 GB of RAM in my computer and a faster processor, I could run a 12 billion parameter model, or even 22 billion parameter model.

You sounded like a financially well off person (or your parents) from some posts you have made before. Which leads me to think you can easily afford buying a much more powerful gaming laptop.

You'll need a good dedicated GPU too (not the Intel one but something like Nvidia or AMD) if your AI can use GPU resources and this definitely means a gaming laptop.

GPU and VRAM can process data A LOT faster this is why Nvidia keeps popping up in AI research. It makes a huge difference.

 

[AlexB23]

You sounded like a financially well off person (or your parents) from some posts you have made before. Which leads me to think you can easily afford buying a much more powerful gaming laptop.

You'll need a good dedicated GPU too (not the Intel one but something like Nvidia or AMD) if your AI can use GPU resources and this definitely means a gaming laptop.

GPU and VRAM can process data A LOT faster this is why Nvidia keeps popping up in AI research. It makes a huge difference.

Yeah, I could, but I am a frugal man. AMD + Vulkan sounds like the way to go, but I am waiting for DDR6 RAM to become standard in the future, as that could boost AI performance as well. Also, I am planning on getting a modular laptop in 2026 or 2027 (Framework), so I can replace parts that die.


DDR6 info:

Leak suggests that DDR6 development has already started, aiming for 21 GT/s

So early?

www.tomshardware.com

Framework Laptop 16 info:

Framework Laptop 16 review: two weeks with the ultimate modular laptop

There’s nothing else like it.

www.theverge.com

 

[Hans Blaster]

Yeah, I understand. It is perfectly okay not to share which field you work in, cos we do not want the Russians or Chinese to take your scientific secrets.

Apparently you are not listening. My work isn't secret or even export controlled. The Russians and Chinese can read it in the journals if they like. (And I know that they do.) I'm just not telling *you* guys. In case you haven't figured it out, "Hans Blaster" isn't my name at all.

 

[AlexB23]

Apparently you are not listening. My work isn't secret or even export controlled. The Russians and Chinese can read it in the journals if they like. (And I know that they do.) I'm just not telling *you* guys. In case you haven't figured it out, "Hans Blaster" isn't my name at all.

I am just joking with you a bit. Alex is not my real name either, but it is based off of my actual name.

 

[timewerx]

Yeah, I could, but I am a frugal man. AMD + Vulkan sounds like the way to go, but I am waiting for DDR6 RAM to become standard in the future, as that could boost AI performance as well. Also, I am planning on getting a modular laptop in 2026 or 2027 (Framework), so I can replace parts that die.


DDR6 info:

Leak suggests that DDR6 development has already started, aiming for 21 GT/s

So early?

www.tomshardware.com

Framework Laptop 16 info:

Framework Laptop 16 review: two weeks with the ultimate modular laptop

There’s nothing else like it.

www.theverge.com

If DDR6 is only less 30% faster in read and write speeds, it's worth the gamble to upgrade now.

 

[AlexB23]

If DDR6 is only less 30% faster in read and write speeds, it's worth the gamble to upgrade now.

It probably is worth the gamble to upgrade now, but I wanna run my computer into the ground first. When it fails to keep a charge unplugged for more than a few minutes, that is when I will replace my laptop.

 

[timewerx]

It probably is worth the gamble to upgrade now, but I wanna run my computer into the ground first. When it fails to keep a charge unplugged for more than a few minutes, that is when I will replace my laptop.

Thanks for NOT contributing to the consumerist madness!

My laptop is couple years old, donated by a relative and still holds charge for couple of hours while working on big, CPU-heavy spreadsheets with the fan working hard. You'll be waiting for several more years!

Anyway, I know how you appreciate being on topic so I'm giving you a database that if you can scrape it online, it will give your AI cutting edge data on medicine. When cutting edge, it can mean it may still be in the research stage that have either underwent human or animal trials but not yet available commercially nor practiced in mainstream medicine but have at least underwent peer review by qualified professionals.

This is the place I go to when I've exhausted all other medical options available to me. I was prediabetic before and this site helped find solutions and cured the disease without spending any money except my time.

PMC Journal List - PMC

Use the PMC Journal List to browse a list of journals that archive some or all of their articles in PubMed Central.

www.ncbi.nlm.nih.gov

 

[AlexB23]

Thanks for NOT contributing to the consumerist madness!

My laptop is couple years old, donated by a relative and still holds charge for couple of hours while working on big, CPU-heavy spreadsheets with the fan working hard. You'll be waiting for several more years!

Anyway, I know how you appreciate being on topic so I'm giving you a database that if you can scrape it online, it will give your AI cutting edge data on medicine. When cutting edge, it can mean it may still be in the research stage that have either underwent human or animal trials but not yet available commercially nor practiced in mainstream medicine but have at least underwent peer review by qualified professionals.

This is the place I go to when I've exhausted all other medical options available to me. I was prediabetic before and this site helped find solutions and cured the disease without spending any money except my time.

PMC Journal List - PMC

Use the PMC Journal List to browse a list of journals that archive some or all of their articles in PubMed Central.

www.ncbi.nlm.nih.gov

You are welcome. The PMC Journals are not combined in one large book, or multiple volumes, so I might not download them.

 

[timewerx]

You are welcome. The PMC Journals are not combined in one large book, or multiple volumes, so I might not download them.

I figured it might be difficult. It will take a lot of work unless you know someone who can mine the data for you. Somehow I trust these journals more than mainstream sources published by MDs even though many of them are still at experimental stages.

There's always a degree of conflict of interest with mainstream sources mainly to generate more profit for the whole healthcare industry.

 

[AlexB23]

I figured it might be difficult. It will take a lot of work unless you know someone who can mine the data for you. Somehow I trust these journals more than mainstream sources published by MDs even though many of them are still at experimental stages.

There's always a degree of conflict of interest with mainstream sources mainly to generate more profit for the whole healthcare industry.

Agreed.

 

[Hans Blaster]

I figured it might be difficult. It will take a lot of work unless you know someone who can mine the data for you. Somehow I trust these journals more than mainstream sources published by MDs even though many of them are still at experimental stages.

PubMed Central is a database of nearly all scientific journals in biology/medicine. Many (in medicine) *are* published by MDs.

There's always a degree of conflict of interest with mainstream sources mainly to generate more profit for the whole healthcare industry.

The PMC listed journals *ARE* the mainstream sources of scientific research in those fields.

 

[timewerx]

The PMC listed journals *ARE* the mainstream sources of scientific research in those fields.

*scientific research* vs certified/approved methods.

It's not the same.

An example of recommendation/conclusion concerning drinking water during exercise:

Clinically-approved version: "please do".
PMC journal with no conflict of interest: "there is no clear relationship of exercise performance and drinking water during exercise"

It makes you think and pertaining to the above example, led to me search even deeper and the information I gathered I experimented on myself. Solved a couple of health issues I've had since childhood. Saving me a lot of money than if I just went straight to the doctor. I have many other examples but in order to keep things on topic, I won't.

Certified/approved methods sometimes isn't the best for your health even if they're not telling you to buy anything. Often times, they're only repeating what every youtube experts already know. And if you're not as healthy as you possibly could, you still end up spending more on healthcare.

Journals will tell you everything even they are not recommending anything. But cross reference the studies and you get a clearer picture of what to do.

Just to make myself clear, I'm not recommending not to drink water during exercise. I do but not as much as many does.

 

[Hans Blaster]

*scientific research* vs certified/approved methods.

It's not the same.

I told you what PMC is. I'm not quite sure what you are trying to claim.

An example of recommendation/conclusion concerning drinking water during exercise:

Clinically-approved version: "please do".

Are there not journals of "clinical practice"? (If there are, I'd bet good money they are in PMC.)

PMC journal with no conflict of interest: "there is no clear relationship of exercise performance and drinking water during exercise"

I don't know what this "drinking water" example is you are talking about, but as I said before PMC is a database of published scientific and medical journals. It is not *a* journal. It is not a category of journals other than in the database or not in the database.

It makes you think and pertaining to the above example, led to me search even deeper and the information I gathered I experimented on myself. Solved a couple of health issues I've had since childhood. Saving me a lot of money than if I just went straight to the doctor. I have many other examples but in order to keep things on topic, I won't.

If you are getting your personal health decisions from PMC, you are doing medicine wrong. See a professional.

Certified/approved methods sometimes isn't the best for your health even if they're not telling you to buy anything. Often times, they're only repeating what every youtube experts already know. And if you're not as healthy as you possibly could, you still end up spending more on healthcare.

Journals will tell you everything even they are not recommending anything. But cross reference the studies and you get a clearer picture of what to do.

This statement makes no sense. It makes me think you don't have any real conception of the materials you are reading and what their purpose is.

Just to make myself clear, I'm not recommending not to drink water during exercise. I do but not as much as many does.

I'm just not interested in "when to drink water". I don't think this thread is about it.

 

[AlexB23]

I told you what PMC is. I'm not quite sure what you are trying to claim.


Are there not journals of "clinical practice"? (If there are, I'd bet good money they are in PMC.)

I don't know what this "drinking water" example is you are talking about, but as I said before PMC is a database of published scientific and medical journals. It is not *a* journal. It is not a category of journals other than in the database or not in the database.

If you are getting your personal health decisions from PMC, you are doing medicine wrong. See a professional.



This statement makes no sense. It makes me think you don't have any real conception of the materials you are reading and what their purpose is.

I'm just not interested in "when to drink water". I don't think this thread is about it.

Be careful my friend. @timewerx is a good guy, but he likes to go on tangents just as I do, though he enjoys non-conventional/unorthodox science. For myself, I prefer sticking to trusted science that is peer-reviewed, or at least not pseudoscience.

 

[timewerx]

Are there not journals of "clinical practice"? (If there are, I'd bet good money they are in PMC.)

I'm sure there are.

But there's also lots of research that are cutting edge though not yet clinically approved which I'm more interested at.

If you are getting your personal health decisions from PMC, you are doing medicine wrong. See a professional.

Probably, but I'm poor. Uninsured and can't afford medicine. Paying off a debt for ten more years. Running a very tight ship and many of my things are donated by relatives but still don't have enough to be able to eat enough food so I'm a little underweight sometimes.

We do have government clinics. Consultations are free but the medicines aren't and we have one of the most expensive medicine price in the region, not to mention, the queues are very long.

So I have to find alternatives. One that doesn't involve money nor begging anyone for free medicines.

 

[timewerx]

Be careful my friend. @timewerx is a good guy, but he likes to go on tangents just as I do, though he enjoys non-conventional/unorthodox science. For myself, I prefer sticking to trusted science that is peer-reviewed, or at least not pseudoscience.

Thanks Alex.

PMC is legit. And if potentially fudged papers are present, the site will indicate to readers conflicts of interest. I don't waste my time on journals with conflicts of interest. And if you wish to take it even further, you can even look up the authors of the journal and see if they're legit too.

MD's don't have all the answers to all disease. They might know about a cutting-edge but non-clinically approved method but they're not going to risk their career to tell you about it.

If all doctors are telling you have an incurable disease and you only have 2 years to live at most, it offers hope there may still be information out there that can save you. Even if it's still at the experimental stage. If you're desperate, it might be worth the risk and better than doing nothing.

 

[Hans Blaster]

Be careful my friend. @timewerx is a good guy, but he likes to go on tangents just as I do, though he enjoys non-conventional/unorthodox science. For myself, I prefer sticking to trusted science that is peer-reviewed, or at least not pseudoscience.

Be careful about what? I can only deal with what is put into posts. When it make no sense, I don't pretend otherwise.

 

[AlexB23]

Be careful about what? I can only deal with what is put into posts. When it make no sense, I don't pretend otherwise.

I am just trying to be a nice guy. My guy @timewerx discussed a pseudo-scientific faster-than-light drive, so it is best that we just trust in articles that are peer reviewed.

By the way, if you want to join my sci-fi thread, feel free to. Fictional FTL tech is discussed, plus a better alternative to the Star Trek warp speed scale is discussed:

Graviton Displacement Drive

Hello folks. I came up with this cool sci-fi drive system for spacecraft. It will not work in real life, but I want to use it in a sci-fi plot. I asked an AI to explain it. Prompt: Explain how the fictional graviton displacement drive works on faster-than-light spacecraft, and how it...

www.christianforums.com