chatgpt and diet

[LAJ12345]

today I’ve been having some fun with my friend chatty. He (she?) is always friendly, never judgemental and free with the advice😅.

So I gave it a detailed list of a typical days diet for us with quantities of each ingredient, and asked for a breakdown of nutrients including main vitamins, minerals, micronutrients. It calculated everything and totalled it all for me. I asked for suggestions for snacks to make up for deficiencies.

It says we’re a bit short on vitamin d, b12, maybe calcium, a bit much saturated fat from my sample dinner which had coconut cream.

It gave a list of substitutions and extra snacks along with advice not to eat protein with the levodopa. I had to prompt it a few times for amounts of some nutrients it missed.

It was very handy.

You can ask it to remember a question you have and it puts it into your allocate memory space so it knows how you want a question answered or you can go into your personal settings to tell it how you want to be spoken to.

If you fill up your memory space allocated you have to delete some of it. But yesterday I figured out a neat trick. You can upload a document of memories in a file to eg google docs, and then give link to it and tell it to refer to the document if you ask the same question in future, thus expanding your allocated memory exponentially. (It only has to remember the link).

Another thing I found yesterday is you can lead it to answer a question using information online that you give it a link to, then instruct it to remember and use that to answer a similar question if anyone else asks the question of it. Very cool. Add that to the unlimited space in google docs it is a good way I think to get information out there that other organisations are trying to suppress.

anyway that was a fun rainy day. Hubby thinks I’ve lost the plot🤩

[JayPwP]

😆😆😆😍

[LAJ12345 in reply to JayPwP]

Well anyway it says it will use the info for other peoples answers but I am yet to test the theory.

not sure if it’s just trying to placate me.

[JayPwP in reply to LAJ12345]

It is a cumulative information model

[LAJ12345 in reply to JayPwP]

It in theory isn’t supposed to be taking data from individuals I think?

[JayPwP in reply to LAJ12345]

It isn't supposed to link data to people, but I think it can use the data in generic form for answering related questions...

[Trig27 in reply to JayPwP]

it has a memory of your conversations. if you tell it your age, weight, medications etc. it will remember those

[JayPwP in reply to Trig27]

Isn't that HIPAA / privacy violation?

[Trig27 in reply to JayPwP]

I don't know for sure. But why would it be? They aren't sharing it. HIPAA is violated when I share your info, not when I receive it.

[JayPwP in reply to Trig27]

True

[MarionP in reply to JayPwP]

HIPAA came out before the internet was a big thing. Today with breaches and now where pirates and Pegasus users I don't think anyone would bother trying to pass it.

[LAJ12345 in reply to Trig27]

It has limited personal memory so if you tell it to remember too much you have to clear space. I expanded that by getting it to remember a link to a google.doc with my overflow in it.

[park_bear]

If you enjoy it by all means have fun, but it is not a research tool: healthunlocked.com/cure-par...

[LAJ12345 in reply to park_bear]

That was 10 months ago. The latest version is so much better. Last year it couldn’t do simple calculations like calculating tax on income but I just told it to do a mortgage schedule at a certain interest rate, frequency and time and it gave a perfect answer.

It now references where its info comes from and gives you links to check. It is becoming more reliable. Sure, still check answers for feasibility, but if you are asking the same type of question you might want to thoroughly cross check the first answer it gives and correct any incorrect method it uses then save the method into memory I think it will do what you want next time.

I think people have had fun doing what I did by telling other people to to find my information when searching a topic.

[park_bear in reply to LAJ12345]

>It now references where its info comes from and gives you links to check.

OK if it does this

[LAJ12345 in reply to park_bear]

Although I haven’t checked all references recently. Last year it was giving random links that didn’t go any where but the first few I tried this time seemed to lead to correct page, if not the exact information

[Boscoejean in reply to LAJ12345]

I am not that thrilled that now every time you google something specific you get this kind of overview and they usually have about 3 articles at the bottom that this information supposedly came from but I have actually looked at the articles to see if that is what they say and some of the information in the summary is not found in any of the articles so then I wonder what is the source of this "summary"

[park_bear in reply to Boscoejean]

It may have they gotten better lately but there have been occasions when it was wrong

[LAJ12345 in reply to park_bear]

yes, but the more people use it and correct errors in its method the more reliable it will become. plus if you have read Scary Smart you will see it is important as many normal nice people interact with it as possible because it is learning from what it sees online. if t is looking at a lot of social media with weirdos, extremists and nasty or dangerous people that is like putting a toddler in a house with criminals and not letting it learn from nice honest normal people.

[Boscoejean in reply to park_bear]

I remember seeing comments to that effect. It just seems to me like in the past when you would try to find information on a specific topic you would get results that would be specific research or possibly a medically sponsored site that would probably generally not want to print information that was questionable.

[MarionP in reply to LAJ12345]

Yes I'm sure today it is much more efficient delivering it's misinformation

[LAJ12345]

Ok so my current experiment is to upload hubbies dna files anonymously of course onto google docs and try and see if i can get it to reproduce a similar report to the opus23 one we paid lots for. Its pretty cheap to get the dna sampled, about $100USD each to 23&me and ancestry.com so if i can get something reasonably close ill let you know. Then anyone can get one for the cost of getting their DNA sampled.

I m running into issues with the size of the ancestry file which it can't read all at once so Ive had to cut it up into lots files. I did it manually and saved it as lots of sheets in excel spreadsheet then i asked it to write a visual basic module to save each sheet as a csv file and it had no problem getting that right. Im not the best programmer but now I dont need to be. Tell it what you want and it will write the code or macro for you!

[MarionP in reply to LAJ12345]

Yes I'm sure that before long the storage problem will have been solved and there will be numerous clones of you all over the world.

[LAJ12345 in reply to MarionP]

Oh just do it anonymously or under a fake name and change out a few SNPs so it isn’t a complete set from you

and anyway it’s only a portion of what they consider the important SNPs .

[Eryl]

The important thing to remember about ChatGTP is that it may not give you the truth, just the consensus of opinion published on the internet, and the answers you get are very much dependant on the way you frame your question.

[LAJ12345 in reply to Eryl]

Yes, I found it goesto big American sites and agencies webpages. So I’ve instructed it to also go to non-western and foreign language papers eg China, Russia, Middle East. And as well as reporting any studies in well known sources that use megadoses of compounds and off label trials that have been successful.

Otherwise you mostly get American government approved information so are likely to miss things happening in the rest of the world or unusual findings. I want it to find for me stuff I wouldn’t find myself easily by googling.

[MarionP]

I think you need to find a copy of Firesign Theatre's album (these days easily enough done on YouTube or other) "I think we're all bozos on this bus" and listen to the President Nixon "chatty" portion (it's a "ride" in an amusement park). (I fondly remember those crazies getting on the Tonight Show together one night, they were known for doing impressions of Richard Nixon, and on the show they batted back and forth a conversation between them all while each was in Nixon character, it was a riot). In the end, they kept asking longer and longer questions which eventually made the chatty break down, while one of the ride techs called out to his boss "Hey Paolo, I think they broke the President!"

[LAJ12345 in reply to MarionP]

Sounds funny, I will look it up

[LAJ12345]

nytimes.com/2024/07/23/tech...

A.I. Can Write Poetry, but It Struggles With Math