• Guest - w'd love to know what you think about the forum! Take the 2025 Survey »

chatgpt diabetes diet?

What does ChatGPT say. Is there a differentiation made between different types of diabetes?
 
I find how a bot replies can be non specific to the question, generic & somewhat alarmist in some cases? In a chirpy”informative” manner..

lol, I once enquired on a website after a purchase where an item that was said to be delivered was & an instant bot reply suggested the police were called.. it turned out to be merely left outside a house 3 doors down..
When I asked for a currier’s account with photographic evidence.. ;)
 
I did a quick Google and it appears the advice is along the lines of Eatwell guide. So no better or worse than the NHS guidelines I was given years ago. That's the diet that helped me on my way to T2. For me wholemeal carbs are as bad as white ones as far as T2 is concerned.
 
The problem is that the "Artificial Intelligence" is not actually intelligent. It scrapes information from all across the internet and combines it in a readable form. Think of it more of a quick summary of thousands of Google hits. As with everything else, garbage in, garbage out. It will get some things right and some things wrong, but will insist it's correct, and can appear very persuasive, particularly if you don't know much about the subject in question.

I've asked an AI a few questions about factual things I am very familiar with and it's shockingly bad at providing correct information.
 
KennyA said:
The problem is that the "Artificial Intelligence" is not actually intelligent. It scrapes information from all across the internet and combines it in a readable form. Think of it more of a quick summary of thousands of Google hits. As with everything else, garbage in, garbage out. It will get some things right and some things wrong, but will insist it's correct, and can appear very persuasive, particularly if you don't know much about the subject in question.

I've asked an AI a few questions about factual things I am very familiar with and it's shockingly bad at providing correct information.
Click to expand...
you can ask ai exactly the same question at different points of the day and over the month can receive differing replies also.

100% agree garbage in, garbage out. ai is only as good as 1) what data actually fed into the ai in the first instance 2) the ability of the programmers who designed the LLM. 3) how the user actually asks questions if at all there is any 'weight' given to the context in which the question was asked which is ofc there always will be to at least some extent. AI is designed with some randomness inbuilt.
 
grantg said:
you can ask ai exactly the same question at different points of the day and over the month can receive differing replies also.

100% agree garbage in, garbage out. ai is only as good as 1) what data actually fed into the ai in the first instance 2) the ability of the programmers who designed the LLM. 3) how the user actually asks questions if at all there is any 'weight' given to the context in which the question was asked which is ofc there always will be to at least some extent. AI is designed with some randomness inbuilt.
Click to expand...

I agree somewhat with all the opinions above. AI is in its infancy but I find I can use it as an additional source of information

Prior to seeing my endo I entered all my test results and asked “ if you were a doctor what actions would you take “.

Ai suggested a prescription very slightly above that of the endo.

As a sole source of information any one source of information can be biased. Luckily for us we live in the 21st century where trust can be established by verification
 
Back
Top