I keep saying medicine will be the first to benefit from AI if its allowed.
Surely, I was jesting
No single team of doctors can harness all the data of a given scenario at once, much less at near immediate response times. I can imagine a desolate location with few modern resources and a small group of people living there, with their own language. Somebody's sick... A user manually cranks some kinda iPad to get some juice and fire it up. AI could take given knowledge, add locale, current events, history of those peoples, etc and suggest a remedy. Then send a drone to deliver the medicine.
This won't be some remote location obscurity - this will replace entire systems.
Hopefully it would drive costs downs but leave the overall quality of public health high.
1) That is not exactly how it works. AI enables machines to learn from data and recognize patterns to perform tasks more efficiently and effectively. The programmer does not have to be an expert in a given field to program algorithms and models to benefit said field.
I wonder how much day to day data is collected. I refuse to go anywhere other than the local Medical University. One of the requirements you must give is your medical data is used anonymously for study. I realize some folks are Leary of insurance discrimination and the like, but if my data helps them find better cures, then I am all for it.
AI is not the database for vehicle maintenance
Well said and it is at the cusp of revolutionizing medical care
This is a real problem with current learning models, but that doesn't mean they won't be able to fix it
Why I said it would need to be a purpose built model. No one is talking about using some current large language model to practice medicine.
But you want other, peripheral information to contribute to the results. The beauty of AI is the ability to recognize patterns far beyond human capability. This is the basis for learning.
Yes! RE my earlier post about detecting pancreatic cancer from seemingly unrelated search terms.
That would need to be on the research end. Researchers could maybe identify signals or even lifestyle treatments, but of course the treatment would need to go through some sort of verification or study. Its actually done today - researchers search through unrelated studies to try to identify trends with other beneficial uses for treatments other than their original design. Presumably AI would be better at this also.
Some of the clinical trials that I support are our drug vs standard of care. There is a huge amount of data collected. AI can help analyze it and recognize any trends, good or bad.That would need to be on the research end. Researchers could maybe identify signals or even lifestyle treatments, but of course the treatment would need to go through some sort of verification or study. Its actually done today - researchers search through unrelated studies to try to identify trends with other beneficial uses for treatments other than their original design. Presumably AI would be better at this also.
Rather than reading it on Reddit
It would be the difference in going to the Dr. and being prescribed a accepted treatment, vs being part of some sort of clinical trial. There related but different.
If only it would do that instead of sometimes providing a false answer. Completely negates its strengths when it fabricates answers.
The problem is that the fix would result in the answer 'we are unable to complete your request'. Which I would personally be okay with, but the current AI models would rather fabricate an answer.
