using AI to do the work for you

[pellicle]

Hi

I've been keeping up with AI developments consistently over the years and the latest stuff (GPT3 and newer) seems to be "better" ... however there exists plenty on YouTube showing that basically it lies. Not only does the question matter (such as which is better vs is this better than that) but you need to interrogate it.

Recently someone messaged me and asked that I look into a specific one, this one:

so I used a SPAM Bucket email account and signed in and asked a basic question in a neutral way on a subject I already knew well

The answer sort of rang true but the "lower bleeding risks" was just not right in my view. I have never seen any article say that; indeed the opposite is written.

So I decided to look for European Heart Journal in 2017 and find anythign written by S. Head et al.

Nope ... so the answer is "it just made it up and made it look good.

So to prevent people Assuming that AI is "the research tool of choice" I want to remind people before you make an Ass out of U (mmmm) to double check the answers, or you'll be D. Head et all not Sam (or whoever he was).

So this goes to show that IF you are trying to reach an informed decision, that you (and you means you) need to get your actual head around the subject, you need to ask critical questions and then you probably need to ask here for views on that point and require us to be able to support with evidence our opinion.

If you can't or don't do the above then you're probably not informing yourself , you're probably just seeking supporting opinions to make you feel better about your personal bias.


PS: if you can find the article where this is taken from please link it below and let me know who the Et Al was

 

[Protimenow]

AI has killed freelance for many writers, myself included Some magazines have even stopped using ANY freelance writers and, I believe, have even fired staff writers and editor, instead using AI for 'writing,' and in some cases using made up names (and photos) of authors. The articles use information taken from copyrighted materials and extracts it, using AI to somehow put together a coherent article

The example given by Pellicle is just the tip of the iceberg. AI isn't entirely trustworthy -- I sure wouldn't give full trust to anything 'written' using AI.

I've lost work to AI and to Grammarly - an app that points out grammatical errors (I used to edit reports that were written in India), ignoring such things as errors in content, erroneous conclusions, and other things that should be used in high priced reports that subscribers are paying for,

Greedy companies that look at a false bottom line, valuing it against accuracy and credibility may eventually learn the lesson about the false economy of using AI to replace its writers. As Pellicle pointed out, conclusions created by AI can't always be trusted and should be approached with skepticism.

 

[pellicle]

The example given by Pellicle is just the tip of the iceberg. AI isn't entirely trustworthy -- I sure wouldn't give full trust to anything 'written' using AI.

indeed, and the best use of it is (assuming you are starting from a position of being informed) to use it as an aggregation of articles for your to read. But you need to read the original source yourself.

Strangely this is exactly what Google has done for at least a decade now, but is actually doing worse now because its AI powered. I believe that Google is now only showing you what its AI thinks you want to see ... to make you happy ... to make you feel you were right all along.

As always tools can act as levers, but its how you pull the lever and where you place the fulcrum that makes the difference.

Best Wishes

 

[AZ Don]

As a former IT worker, many of my associates were critical of the work done when outsourced to India. My concern was that the work done in India was actually quite good, and so jobs would keep being outsourced. Still it wasn't really Apples to Apples because my company would target older, experienced employees for layoff and they would be replaced by young, less experienced workers in India. I feel similarly with AI. I don't disagree with anything stated above. AI absolutely makes mistakes and makes things up, though sometimes people do too. But AI does some things surprisingly well, and like the power of computer chips, it is expected to improve exponentially. Just compare the answer from AI to an older technology such as Alexa ... that barely knows how to do a search. Or look at what Tesla has done. I tested their self driving technology last May. It reminded me of a young inexperienced driver - I absolutely would not trust it other than in testing. But it did not make a mistake, it just wasn't a smooth driving experience. I expect a very different experience within a few years.

 

[kentuckyed]

AI and social media in general appears to be a driving force in attempting to raise the average individual to a higher plane of understanding of just about every topic. The problem I see is that after just a cursory review of a subject that individual thinks they're now an expert and goes about acting as such. I'm not sure this is really beneficial at all. Sometimes I think that the average man has advanced just about as far as he's going to go for quite a bit.

Now I'll try a bit of Pelliclle magic, hope it works

 

[pellicle]

The problem I see is that after just a cursory review of a subject that individual thinks they're now an expert and goes about acting as such

Good point
AI Reinforced Dunning Kruger

 

[perrybucsdad]

PS: if you can find the article where this is taken from please link it below and let me know who the Et Al was

I believe this is Stuart J Head, and the article is titled "Mechanical versus bioprosthetic aortic valve replacement" in volume 38 issue 28 July 21 2017.. pp 2183 - 2191.

There were a number of citations for Head in 2017, but I think that was the one it may have come from. I haven't read the document so not sure if it interpreted it correctly or not.

 

[pellicle]

I believe this is Stuart J Head, and the article is titled "Mechanical versus bioprosthetic aortic valve replacement" in volume 38 issue 28 July 21 2017.. pp 2183 - 2191.

thanks for that

As expected Stuart doesn't say less bleeding risk either ... so the AI just "made **** up" on that point (which was my point). Personally I don't think you'll find any surgeon or cardiologist tell you that "the risks of bleeding are less" with a mechanical valve and ACT ... to my mind they are only likely to go as far as saying "well controlled ACT will give you about the age related general population risk of bleeding complications"

https://pubmed.ncbi.nlm.nih.gov/28444168/

Abstract​

Mechanical valves used for aortic valve replacement (AVR) continue to be associated with bleeding risks because of anticoagulation therapy, while bioprosthetic valves are at risk of structural valve deterioration requiring reoperation. This risk/benefit ratio of mechanical and bioprosthetic valves has led American and European guidelines on valvular heart disease to be consistent in recommending the use of mechanical prostheses in patients younger than 60 years of age.
Despite these recommendations, the use of bioprosthetic valves has significantly increased over the last decades in all age groups. A systematic review of manuscripts applying propensity-matching or multivariable analysis to compare the usage of mechanical vs. bioprosthetic valves found either similar outcomes between the two types of valves or favourable outcomes with mechanical prostheses, particularly in younger patients. The risk/benefit ratio and choice of valves will be impacted by developments in valve designs, anticoagulation therapy, reducing the required international normalized ratio, and transcatheter and minimally invasive procedures. However, there is currently no evidence to support lowering the age threshold for implanting a bioprosthesis. Physicians in the Heart Team and patients should be cautious in pursuing more bioprosthetic valve use until its benefit is clearly proven in middle-aged patients.