AI - Why you should be worried about this

mikey
mikey
Joined: 22 Jan 05
Posts: 12654
Credit: 1839045286
RAC: 4645

China develops new chip due

China develops new chip due to import restrictions, I wonder if it could crunch?

https://www.yahoo.com/tech/specialist-carbon-nanotube-ai-chip-093020487.html

B.I.G
B.I.G
Joined: 26 Oct 07
Posts: 117
Credit: 1161578685
RAC: 975395

I'll throw in a different

I'll throw in a different opinion for discussion. And yes I am biased as all of us so be welcome to counter my arguments :) I can't promise to reply because of time tough...

Edit: oh well... are we on the same page reading your argument again... maybe I just in detail described what you say... not have a machine making the final verdict but up to that do everything to support the human?

 

mikey wrote:

I also think Medicine should not be on the bleeding edge of AI either!! I DO think that the internet in general can help Doctors diagnose things they have never or rarely seen before giving patients the care they need in a timely matter. BUT letting a computer diagnose and then suggest the proper medicines to take is far from DEPENDING on a computer, or piece of software ie AI, to SOLELY do it ALL is dangerous and all medical organizations should highly regulate the use of it as such.

Before my work got me in touch with the medical world I might have thought the same... now I wish we had an algorithm going trough sites like pubmed (and also is capable of accessing information behind the pay wall) and then making a estimated diagnosis out of that because in most countries of the world healthcare is in a very bad state and does not follow the science, neither in diagnosis, nor in treatment. It's not that doctors do this on purpose, they are either ignorant about educating themselves better or just have no time to do so. And on top they are incapable of interpreting medical studies because that's something they never learned in their education. So they get some nice info folder from a pharma company and they believe all of that because pharma companies make medicine and that's good, right? I have seen this so often with friends, the get diagnosed, then the patient doesn't react as expected by the doctor, the doctor is confused and suggests to try whatever so they can see what happens and eventually figure out how the patient is off the least worse. Not because they are bad people, they just don't  even know where and how to get information and have no time at all, but are expected to know it all. They are in a horrible spot and it takes a huge burden on many.

But even if you made doctors use scientific sites like pubmed they couldn't digest the information there. Something that was learned the hard way during the pandemic. Scientists published new knowledge like crazy but hardly anybody had the ability to read it. It was misinterpreted and in consequence very wrong decisions were made. We all felt how it turned out but few know how it turned out that way and why.

Then we have big companies manipulating the publications to not harm their profits, but often poor countries who can't afford these products anyway are quite neutral in their research, some governments spend money on research independent from pharma companies, that creates a huge mess of information. The beauty of science however is: you can't hide anything. If a study doesn't publish the principle and limitations you already know they hide something. And if you read how the study was set up you see the limitations and can search for additional information on that. So if we have technological support in doing that it would be great.

Of course, the algorithm can be biased too, and issue that was often found, humans program it with a bias or the data that is used contains a bias and an algorithm is only as good as it's source.

In general doctors constantly misdiagnose and in consequence mistreat their patients because of a lack of knowledge and time and bias in the medical field. The human body is too complex for our medical system. Single private doctors who work half a month and then educate themselves half a month and write you a bill accordingly can avoid that, but not the majority.

Please let us have an algorithm that can crawl trough all those studies plus addition information like personal experience and group them by how they are relevant to each other, highlight the differences - sound a loud alarm at differences, give a quick overview with direct links to the original part of the text.

It was an algorithm that figured out that 2 very popular medications sold in the US do interact with each other and must not be taken together. How? Well... people searched for those 2 with the added word "interactions" that caused a spike in the database and no human had ever found that. So yes, please let an algorithm run trough all scientific and non scientific sites before giving patients medicine, or make a diagnosis because symptoms of disease and malnourishment or wrong nutrients might be the same in 30 out of 35 indicators and they change constantly as new knowledge arrives. And have a big red light flash that says "contact a scientist before further action". The doctor knows the patient, the scientist knows how to handle the data, they have to talk with each other. And they need support in doing so.

It has to be machine learning and algorithms because knowledge is changing so fast, so much new information comes out every year. You thought you are educated about something, 3 months later all has changed because they found more information, did 20 studies around the whole world about it and to read all those takes months. If you are not a scientist specialised to a very specific field it's impossible to follow, yet alone when you treat patients during the day. We need something that helps with diagnosis and treatment.

Today we say it takes 20 years from scientific proof until knowledge spreads to the hospitals, there is no need for that to take that long. Have an algorithm flash a red sign in every hospital that says "new knowledge", employ a scientist who is trained to interpret studies at every hospital and have him discuss this with the doctors.

But, and this is the most important: it always has to be a human that does the final verification, and seeing how doctors work today... yes those who warn that doctors just would use AI for diagnose could be right. But how else to handle the insane amount of knowledge?

I am absolutely not afraid of applied mathematics and I think nobody would be... it's just that AI term people are afraid of.

Fun fact, there is already so much machine generated content on the internet that tools like chatGPT run into problems because they start to recycle their own texts and that doesn't work. So eventually the whole thing breaks itself. To avoid this companies spend huge money to preserve archives of the internet at a time before those algorithms started generating content and then use those archives for content creation.

 

So that thing that they call AI is just another technology that will bring us great opportunities and equally great challenges, like mobile phones and the internet did, like contraceptives did - we still haven't figured out how to handle the consequences of any of these technologies fully and now already the next huge one is there.

I'm not worried about a single technology, I'm worried we outpace ourself at a speed our society can't adjust to. But there is no way we artificially can slow this down, we just can try our hardest to face and overcome that challenges. The beauty of life is that these things get automatically corrected - societies that stop to function - for whatever reason - collapse and with that make space for something new that will look at the lessons. That would be the natural slow down. We have seen this happening all the time trough human history.

mikey
mikey
Joined: 22 Jan 05
Posts: 12654
Credit: 1839045286
RAC: 4645

B.I.G wrote: I'll throw in a

B.I.G wrote:

I'll throw in a different opinion for discussion. And yes I am biased as all of us so be welcome to counter my arguments :) I can't promise to reply because of time tough...

Edit: oh well... are we on the same page reading your argument again... maybe I just in detail described what you say... not have a machine making the final verdict but up to that do everything to support the human?

 

mikey wrote:

I also think Medicine should not be on the bleeding edge of AI either!! I DO think that the internet in general can help Doctors diagnose things they have never or rarely seen before giving patients the care they need in a timely matter. BUT letting a computer diagnose and then suggest the proper medicines to take is far from DEPENDING on a computer, or piece of software ie AI, to SOLELY do it ALL is dangerous and all medical organizations should highly regulate the use of it as such.

Before my work got me in touch with the medical world I might have thought the same... now I wish we had an algorithm going trough sites like pubmed (and also is capable of accessing information behind the pay wall) and then making a estimated diagnosis out of that because in most countries of the world healthcare is in a very bad state and does not follow the science, neither in diagnosis, nor in treatment. It's not that doctors do this on purpose, they are either ignorant about educating themselves better or just have no time to do so. And on top they are incapable of interpreting medical studies because that's something they never learned in their education. So they get some nice info folder from a pharma company and they believe all of that because pharma companies make medicine and that's good, right? I have seen this so often with friends, the get diagnosed, then the patient doesn't react as expected by the doctor, the doctor is confused and suggests to try whatever so they can see what happens and eventually figure out how the patient is off the least worse. Not because they are bad people, they just don't  even know where and how to get information and have no time at all, but are expected to know it all. They are in a horrible spot and it takes a huge burden on many.

But even if you made doctors use scientific sites like pubmed they couldn't digest the information there. Something that was learned the hard way during the pandemic. Scientists published new knowledge like crazy but hardly anybody had the ability to read it. It was misinterpreted and in consequence very wrong decisions were made. We all felt how it turned out but few know how it turned out that way and why.

Then we have big companies manipulating the publications to not harm their profits, but often poor countries who can't afford these products anyway are quite neutral in their research, some governments spend money on research independent from pharma companies, that creates a huge mess of information. The beauty of science however is: you can't hide anything. If a study doesn't publish the principle and limitations you already know they hide something. And if you read how the study was set up you see the limitations and can search for additional information on that. So if we have technological support in doing that it would be great.

Of course, the algorithm can be biased too, and issue that was often found, humans program it with a bias or the data that is used contains a bias and an algorithm is only as good as it's source.

In general doctors constantly misdiagnose and in consequence mistreat their patients because of a lack of knowledge and time and bias in the medical field. The human body is too complex for our medical system. Single private doctors who work half a month and then educate themselves half a month and write you a bill accordingly can avoid that, but not the majority.

Please let us have an algorithm that can crawl trough all those studies plus addition information like personal experience and group them by how they are relevant to each other, highlight the differences - sound a loud alarm at differences, give a quick overview with direct links to the original part of the text.

It was an algorithm that figured out that 2 very popular medications sold in the US do interact with each other and must not be taken together. How? Well... people searched for those 2 with the added word "interactions" that caused a spike in the database and no human had ever found that. So yes, please let an algorithm run trough all scientific and non scientific sites before giving patients medicine, or make a diagnosis because symptoms of disease and malnourishment or wrong nutrients might be the same in 30 out of 35 indicators and they change constantly as new knowledge arrives. And have a big red light flash that says "contact a scientist before further action". The doctor knows the patient, the scientist knows how to handle the data, they have to talk with each other. And they need support in doing so.

It has to be machine learning and algorithms because knowledge is changing so fast, so much new information comes out every year. You thought you are educated about something, 3 months later all has changed because they found more information, did 20 studies around the whole world about it and to read all those takes months. If you are not a scientist specialised to a very specific field it's impossible to follow, yet alone when you treat patients during the day. We need something that helps with diagnosis and treatment.

Today we say it takes 20 years from scientific proof until knowledge spreads to the hospitals, there is no need for that to take that long. Have an algorithm flash a red sign in every hospital that says "new knowledge", employ a scientist who is trained to interpret studies at every hospital and have him discuss this with the doctors.

But, and this is the most important: it always has to be a human that does the final verification, and seeing how doctors work today... yes those who warn that doctors just would use AI for diagnose could be right. But how else to handle the insane amount of knowledge?

I am absolutely not afraid of applied mathematics and I think nobody would be... it's just that AI term people are afraid of.

Fun fact, there is already so much machine generated content on the internet that tools like chatGPT run into problems because they start to recycle their own texts and that doesn't work. So eventually the whole thing breaks itself. To avoid this companies spend huge money to preserve archives of the internet at a time before those algorithms started generating content and then use those archives for content creation.

 

So that thing that they call AI is just another technology that will bring us great opportunities and equally great challenges, like mobile phones and the internet did, like contraceptives did - we still haven't figured out how to handle the consequences of any of these technologies fully and now already the next huge one is there.

I'm not worried about a single technology, I'm worried we outpace ourself at a speed our society can't adjust to. But there is no way we artificially can slow this down, we just can try our hardest to face and overcome that challenges. The beauty of life is that these things get automatically corrected - societies that stop to function - for whatever reason - collapse and with that make space for something new that will look at the lessons. That would be the natural slow down. We have seen this happening all the time trough human history. 

YES we are thinking along the same lines but I would go one step further to bring in medical studies and even non medical things that may work as well as giving people a pill, ie xyz OTC product because for patients with mild symptoms it's better than an RX. Acupuncture is another thing that can bring relief for some people without the need for any RX meds. There are a host of other things that the internet is full of, MOST, being quacks, but ie Chinese Medicine that has been used for Centuries with great affect. And NO I'm not saying people should start grinding up dead lizards or whatever!!

I had a HUGE rash one time and the US trained Dr put me on ONE kind of antibiotic, an IV twice a day every day, while waiting 10 days for the culture to grow in the Lab. I saw a Chinese trained Dr on a follow-up, the Iv was not making a dent in the rash, and was SHOCKED the original Dr hadn't put me on a multi-spectrum antibiotic pill as well and within 2 days the rash started to disappear!! The problem wasn't that the US trained Dr wasn't competent it's just that his education was limited to what he was taught by other US trained Dr's and that's it. This is an example where AI could have advised him to try this as well and then he could have made the decision to do so or not.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.