my tinfoil hat fish conspiracy.

[OnlyGenusCaps]

On 4/19/2022 at 5:05 PM, Cory said:

I wrote the first paragraph, then used AI to do the rest.

Well that's it.  I'm sold!  I'm going to use AI to start writing all of my forum posts, like all humans in a goodly manner.

[Patrick_G]

This thread has me in stitches! It’s crazy that these site get such high rankings in search. I guess it accounts for all the stupid advice on Facebook aquarium pages. 

[modified lung]

On 4/19/2022 at 4:24 PM, OnlyGenusCaps said:

Well that's it.  I'm sold!  I'm going to use AI to start writing all of my forum posts, like all humans in a goodly manner.

No one will know the difference if you pretend you're not AI. It's too late to kill a child. Some people may have more information than most, but a child may be born prematurely and died because everyone who thought she was your biological mother passed on or had to go through the process of becoming an AI later in life. That's an ongoing nightmare in which a parent, guardian, or partner can get hurt in any given day. The human race can survive. It can be killed in very short periods of time. A lot of the pain that happens in the process of raising a child goes to the father or mother.  

I don't believe in an artificial intelligence. It's a huge misconception that robots are going to run the world until there's enough money to care for the babies, or if they will.

In the meantime you may consider a new approach to AI for your pet. You may also develop a smart cat that would automatically make mistakes in your life.

I've only read two books. You're probably best off taking one if you're a dog. You might try doing something with your dog - you might ask your dog. Doing such research is hard enough and often time consuming. Do I recommend this book to others who don't know a thing about it and like to see a dog in your life the way it is?

[PineSong]

On 4/19/2022 at 7:50 PM, modified lung said:

No one will know the difference if you pretend you're not AI. It's too late to kill a child. Some people may have more information than most, but a child may be born prematurely and died because everyone who thought she was your biological mother passed on or had to go through the process of becoming an AI later in life. That's an ongoing nightmare in which a parent, guardian, or partner can get hurt in any given day. The human race can survive. It can be killed in very short periods of time. A lot of the pain that happens in the process of raising a child goes to the father or mother.  

I don't believe in an artificial intelligence. It's a huge misconception that robots are going to run the world until there's enough money to care for the babies, or if they will.

In the meantime you may consider a new approach to AI for your pet. You may also develop a smart cat that would automatically make mistakes in your life.

I've only read two books. You're probably best off taking one if you're a dog. You might try doing something with your dog - you might ask your dog. Doing such research is hard enough and often time consuming. Do I recommend this book to others who don't know a thing about it and like to see a dog in your life the way it is?

Sadly I’ve read and graded university papers that read like this. Just put “Chaucer” and “The Knight’sTale” in every other sentence and you’ll get a solid C. 

[Brandon p]

On 4/19/2022 at 9:05 PM, PineSong said:

Sadly I’ve read and graded university papers that read like this. Just put “Chaucer” and “The Knight’sTale” in every other sentence and you’ll get a solid C. 

How long ago was that. All are papers ran through a program that would says how much was plagiarized were and if topics or ideas were used and how much %. Private college. 

[Brandon p]

I do use some material and copy and paste.some things. I’m a poor speller so sometimes I just copy and paste scientific names. I try site the Author if if is more than a few lines. I rarely site web pages because I don’t want  want get in any issues will the community. 

[PineSong]

On 4/19/2022 at 10:47 PM, Brandon p said:

How long ago was that. All are papers ran through a program that would says how much was plagiarized were and if topics or ideas were used and how much %. Private college. 

As recently as five years ago and as long ago as 25! Turnitin.com and others can detect plagiarized nonsense but not newly generated nonsense 🙂

[Odd Duck]

On 4/19/2022 at 9:56 PM, Brandon p said:

I do use some material and copy and paste.some things. I’m a poor speller so sometimes I just copy and paste scientific names. I try site the Author if if is more than a few lines. I rarely site web pages because I don’t want  want get in any issues will the community. 

Nothing wrong with copy paste on a name, and if you put a quoted section inside quotation marks and mention it was quoted from another website, you’re safe enough if it’s only a few sentences.

[Torrey]

On 4/19/2022 at 1:00 PM, ndfi78 said:

mmm, fried spam... so delicious! 

On 4/19/2022 at 4:05 PM, Cory said:

Here is a version I made for you guys on the fly. AI copyrighting is all the rage in advertising over the past year.

Cory, thank you so much for the belly laughs and the example. I was going to explain AI, but this is so much better than my explanation, lol

 

On 4/19/2022 at 6:57 AM, drewzero1 said:

There are a lot of sites that make a business of churning out "articles" that appear convincing enough to search engines like Google, but that don't pass the sniff test for a human reader. A lot of the content is automatically generated or paraphrased (and I use the word loosely here) from legitimate sites either by bots or non-native speakers.

I was searching for information on different kinds of sausages and this came up. I'm a bit worried some space aliens are sharing recipes to "bring a different kind of feel to the humans."

Through all of this, the recurring theme in the back of my head was a total Floyd moment....

 

I have run into so much frustration since even some so-called peer-reviewed journals have even been permeated with "peer-review" that is questionable in their credentials for their "peers" who are supposedly doing the review.

I love having more access to information, I guess with the increased access to information we have even more responsibility to verify accuracy?

The most frustrating articles are best explained by Cory's example above: Starts out with accuracy.... and then once trust is established with verifiable information, the AI steps in and illustrates the faults in our logic frames for teaching AI... which is how the AI car ran into a schoolbus that was laying on it's side after being in an accident... because the AI didn't "recognize" the logic frame that explained "schoolbus"

(Apparently that particular article is now behind a paywall, from 2015. This one helps explain some of the issues with AI: https://www.thedrive.com/tech/17618/mercedes-will-teach-its-self-driving-cars-to-recognize-school-buses )

 

 

[Ken Burke]

On 4/18/2022 at 8:08 PM, lefty o said:

use aluminum foil instead.  a lot of articles, wikipedia, etc are just something someone mumbled, that gets repeated over and over. at least here on the forum, if i say something that is bulldung, someone can step in and say so, and there remains a record of it. that way someone comes along down the road can see for themselves that what i said is likely no good, vs. there just being a paragraph of just wrong info with no counterpoint.

Right behind u Big Guy!

how many dumb things can I say in a day? Apparently one more🫣

[Ken Burke]

On 4/19/2022 at 5:05 PM, Cory said:

Here is a version I made for you guys on the fly. AI copyrighting is all the rage in advertising over the past year.

 

I wrote the first paragraph, then used AI to do the rest.  Here is a 4 minute video of how I did it and it works. https://www.loom.com/share/f176419120d04095accf74d404f77330

Aquarium plants are relatively simple. Make sure you have a good light that puts out a plant growing spectrum. Pair this with fertilizers such as Easy Green Liquid and Root tabs. Then wait as your plants convert and grow into a lush aquarium.

Red aquarium plants bring energy to the tank and add a splash of color to the water. They’re also great for mixing up the look of your aquarium. In this article, we’ll look at some of the best red aquarium plants, along with some tips for growing them in your tank.

This can take some time, so be patient. Don’t let your plants run out of oxygen either. For the best results, set up a system where the roots can breath through an air pump. This will increase the health and growth of your aquarium plants.

In time, they’ll add life to your tank and provide a home to all sorts of beautiful fish and creatures. From tiny anemones to large corydoras, plants are the life blood of an aquarium. They keep your tank clean and provide the ecosystem that fish need to live and thrive. To learn more, check out our complete guide to aquarium plants.

Not sure I would recommend this method to the technical writers.  But it does explain some of the proposals I’ve seen coming out of the defense industrial base….

On 4/19/2022 at 8:05 PM, PineSong said:

Sadly I’ve read and graded university papers that read like this. Just put “Chaucer” and “The Knight’sTale” in every other sentence and you’ll get a solid C. 

Is that what I was doing wrong?!  Oh wait, AI wasn’t a thing in the 80’s.

[Ken Burke]

On 4/19/2022 at 9:56 PM, Brandon p said:

I do use some material and copy and paste.some things. I’m a poor speller so sometimes I just copy and paste scientific names. I try site the Author if if is more than a few lines. I rarely site web pages because I don’t want  want get in any issues will the community. 

I can’t spell.  But my vocabulary is a little larger since I have to find words I can spell….😳

[lefty o]

On 4/19/2022 at 10:45 PM, Ken Burke said:

I can’t spell.  But my vocabulary is a little larger since I have to find words I can spell….😳

my vocabulary gets smaller as i get older, and my spelling gets worse.🤪

[Brandon p]

On 4/19/2022 at 10:58 PM, PineSong said:

As recently as five years ago and as long ago as 25! Turnitin.com and others can detect plagiarized nonsense but not newly generated nonsense 🙂

It was not turnitin.com that what they use at my sons  high school. I was about 5 or 6 years ago for my masters. I wish I could remember because it would tell you how simlair it was to other articles and most scholarly journals. It was a problem at times because you could site one paper and that paper was 70% copied from a prior study or Scientific opinion

[freemoney]

On 4/19/2022 at 11:28 PM, Torrey said:

Cory, thank you so much for the belly laughs and the example. I was going to explain AI, but this is so much better than my explanation, lol

 

Through all of this, the recurring theme in the back of my head was a total Floyd moment....

 

I have run into so much frustration since even some so-called peer-reviewed journals have even been permeated with "peer-review" that is questionable in their credentials for their "peers" who are supposedly doing the review.

I love having more access to information, I guess with the increased access to information we have even more responsibility to verify accuracy?

The most frustrating articles are best explained by Cory's example above: Starts out with accuracy.... and then once trust is established with verifiable information, the AI steps in and illustrates the faults in our logic frames for teaching AI... which is how the AI car ran into a schoolbus that was laying on it's side after being in an accident... because the AI didn't "recognize" the logic frame that explained "schoolbus"

(Apparently that particular article is now behind a paywall, from 2015. This one helps explain some of the issues with AI: https://www.thedrive.com/tech/17618/mercedes-will-teach-its-self-driving-cars-to-recognize-school-buses )

 

 

I read a study, a few years back where it was estimated over 40% of peer reviewed articles couldn’t be replicated. There was also 3 academics that went out and made up the most ridiculous articles they could think of, and got quite a few peer reviewed.

  Most of the academics today have, and deserve, no credibility whatsoever.

[CT_]

On 4/19/2022 at 10:16 PM, freemoney said:

  Most of the academics today have, and deserve, no credibility whatsoever.

oof.  as an ivory tower academic I feel compelled to defend my self a little.  Fraud happens, and we'll never know the exact ratios, but I suspect its the minority of the "reproducibility epidemic".  The big article recently (5-10 years ago?) about this was cancer research but in general it is an issue throughout many of the sciences and in bio-related fields especially.  Its also important to remember that headlines like "40% of science can't be reproduced" is probably more accurately "20 out of 50 papers, in one area, chosen by one group, couldn't be reproduced by that group in time to write the paper about how there's a problem with writing papers"

 

I'd say the majority of irreproducible results come from a few places:

1) False positives and positive results publishing bias.  If you require 95% confidence in a result then 1/20 of your results are false positives.  add to that the fact the negative results are less commonly published a you are, maybe counter intuitively, going to see a high rate of false positives (>5%).

2) Bad methods communication.  Often the methods section of a paper just says we did common protocol XYZ with kit A from company B.  But something that possibly the researches didn't even know was critical wasn't communicated, like how something was handled after centrifugation, or the way something was washed, etc. 

3) Mistakes in data analysis.  This is getting better and more and more people are sharing code and data with papers, but you can still find a lot of papers that don't make it easy to get the data or the code.

4) Questionable practices.  It happens.  Like I said above IDK how often, but it does happen.  I do believe its the least common.

BONUS 5) Press.  This isn't really about reproducibility but I think the press distorting findings, misunderstanding, and reporting on reports written by other reporters causes a lot of wrong stuff to be said for the sake of headlines that eventually leads to distrust, especially when that thing the press hyped as true that the scientists never said was true turns out not to be true.

On 4/19/2022 at 10:16 PM, freemoney said:

There was also 3 academics that went out and made up the most ridiculous articles they could think of, and got quite a few peer reviewed.

In Computer science there was a classic "randomly generated paper" that got accepted years ago.  Its been repeated with success many times since.  This is less a problem with scientists and more a problem with predatory journals that publish anything in exchange for a few dollars.  (it costs between a few hundred and a few thousand dollars to publish a scientific paper)

[OnlyGenusCaps]

On 4/19/2022 at 10:28 PM, Torrey said:

I have run into so much frustration since even some so-called peer-reviewed journals have even been permeated with "peer-review" that is questionable in their credentials for their "peers" who are supposedly doing the review.

There are a number of problems with incentives within the peer review system at this time, and I could go on about it, but I suspect that would take us way, way down the garden path.  Suffice it to say that between predatory journals (as @CT_ brought up), now incentives in P&T for reviews done, little value placed on replicating others' results to validate them, and greater weight placed on "productivity" than quality of research, the system needs some help.

On 4/20/2022 at 12:16 AM, freemoney said:

I read a study, a few years back where it was estimated over 40% of peer reviewed articles couldn’t be replicated. There was also 3 academics that went out and made up the most ridiculous articles they could think of, and got quite a few peer reviewed.

CT_ has a great and details explanation above, but I think there are two things that should be added.

Those things have been happening primarily in humanities and social studies fields, and it's a problem.  Those fields know it.  Since the 1970's there have been a growing number of disciplines based around social dogmas.  So, I think it is important to recognize that: academics ≠ scientists

It's true medicine has some issues with this as well, but it's usually from the early stages, and driven by what I think is a good decision: we've decided not to allow researchers to experiment on people, leading to many correlative studies.  This is why eggs are bad for you one day and then good the next, only to be vilified the day after.  Not running highly controlled experiments on people hampers medical advances, but I'm good with not allowing people to scoop others out of their cars to suddenly be taken off to be part of a medical experiment.  Seems like an okay trade off to me. 

On 4/20/2022 at 12:16 AM, freemoney said:

Most of the academics today have, and deserve, no credibility whatsoever.

Flawed as aspects of the system may be, I will always take inquiry over ignorance.  My respect will be directed toward those trying to build knowledge, and never to those who are only capable of tearing it down.  As my dad always told me "It takes care, time, and skill to build something.  But any idiot can destroy stuff."  I am amazed at how often today we hold up those as paragons who only seem to be able to tear down others work. 

[Ken Burke]

On 4/20/2022 at 1:02 AM, CT_ said:

oof.  as an ivory tower academic I feel compelled to defend my self a little.  Fraud happens, and we'll never know the exact ratios, but I suspect its the minority of the "reproducibility epidemic".  The big article recently (5-10 years ago?) about this was cancer research but in general it is an issue throughout many of the sciences and in bio-related fields especially.  Its also important to remember that headlines like "40% of science can't be reproduced" is probably more accurately "20 out of 50 papers, in one area, chosen by one group, couldn't be reproduced by that group in time to write the paper about how there's a problem with writing papers"

 

I'd say the majority of irreproducible results come from a few places:

1) False positives and positive results publishing bias.  If you require 95% confidence in a result then 1/20 of your results are false positives.  add to that the fact the negative results are less commonly published a you are, maybe counter intuitively, going to see a high rate of false positives (>5%).

2) Bad methods communication.  Often the methods section of a paper just says we did common protocol XYZ with kit A from company B.  But something that possibly the researches didn't even know was critical wasn't communicated, like how something was handled after centrifugation, or the way something was washed, etc. 

3) Mistakes in data analysis.  This is getting better and more and more people are sharing code and data with papers, but you can still find a lot of papers that don't make it easy to get the data or the code.

4) Questionable practices.  It happens.  Like I said above IDK how often, but it does happen.  I do believe its the least common.

BONUS 5) Press.  This isn't really about reproducibility but I think the press distorting findings, misunderstanding, and reporting on reports written by other reporters causes a lot of wrong stuff to be said for the sake of headlines that eventually leads to distrust, especially when that thing the press hyped as true that the scientists never said was true turns out not to be true.

In Computer science there was a classic "randomly generated paper" that got accepted years ago.  Its been repeated with success many times since.  This is less a problem with scientists and more a problem with predatory journals that publish anything in exchange for a few dollars.  (it costs between a few hundred and a few thousand dollars to publish a scientific paper)

I remember my thesis advisor lamenting when asked to provide more supporting studies in one of his papers.  Talk about hung dog….

I would add that there is a bias when the research in question goes along with our world view.  That path leads to group think.

 

[JettsPapa]

On 4/19/2022 at 8:05 PM, PineSong said:

Sadly I’ve read and graded university papers that read like this. Just put “Chaucer” and “The Knight’sTale” in every other sentence and you’ll get a solid C. 

I took Driver's Ed in high school back in the day.  The teacher was a coach who did a good teaching the actual driving part, but let's say he was less than diligent with the classroom part.

We had questions at the end of each chapter we had to answer as part of the course, and he told us that when we turned in our work we should answer the questions with complete sentences.  That wasn't a problem, but I wondered why.  I finally figured out that he wasn't reading the questions.  After that my answers would be in complete sentences, and they would be correct, but they didn't always actually answer the question that was asked.  Either he never caught on, or didn't care.

[Guppysnail]

@JettsPapa I think I may have witnessed a few of his students that did not pay or read either 🤣

[lefty o]

On 4/20/2022 at 9:14 AM, Guppysnail said:

@JettsPapa I think I may have witnessed a few of his students that did not pay or read either 🤣

there's a whole lot of 'em out there.😲

[Hobbit]

I’m too late to the party to add much, but I’ll just say that my husband is an academic researcher professor dude in computer science and his research involves AI. Be forewarned that you can have an AI give you articles, recipes, poems, songs, stories that did not exist… and it’s not just text. It can give you Landscapes, cats, people, and all sorts of images that do not exist in real life.

They’re hilarious when they make mistakes, and super creepy when they get it right.

[Widgets]

I will just add that there is all different shades of plagiarism. Back in the day, for the senior project for my BS in Computer Science, I did a project on an AI search algorithm and developed an efficiency tweak to the standard algorithm. My mentor professor continued the work and about a year later sent me a copy of a paper he had published. He had me as a contributing author. About 80% of the paper was lifted verbatim from my senior project paper.

A funny thing about is that immediately upon reading the paper, I knew he did not use the new algorithm to its fullest potential. I called him up and told him he had done it wrong. I explained how he should change his testing process. About a year later he sent me another paper where he had corrected his usage and was showing great results. He again carried me as a contributing author, and about 75% of the paper was still my senior project paper.

An even funnier thing was the closing statement of the paper, summarized that my updated algorithm seemed to work more efficiently the larger the problem space, but had no hypothesis as to why. I pulled out my trusty calculator and in less than 5 minutes had an ironclad explanation of the increase in efficiency. 

That professor tried tirelessly to lure me away from my Aerospace career towards one of academia. 

Edited April 20, 2022 by Widgets
Stupid typos

[Irene]

On 4/19/2022 at 4:05 PM, Cory said:

Here is a version I made for you guys on the fly. AI copyrighting is all the rage in advertising over the past year.

Guess I'm not out of a job yet... 😂

[Tanked]

On 4/20/2022 at 9:10 AM, JettsPapa said:

I took Driver's Ed in high school back in the day.  The teacher was a coach who did a good teaching the actual driving part, but let's say he was less than diligent with the classroom part.

We had questions at the end of each chapter we had to answer as part of the course, and he told us that when we turned in our work we should answer the questions with complete sentences.  That wasn't a problem, but I wondered why.  I finally figured out that he wasn't reading the questions.  After that my answers would be in complete sentences, and they would be correct, but they didn't always actually answer the question that was asked.  Either he never caught on, or didn't care.

I learned that one of my high school teachers took most of the test questions from the first 2 or 3 paragraphs. I still read all of the material, but only needed to review the first few paragraphs.

Later as an adult taking some night classes the teacher told the class. "You don't have to remember all of this stuff, you just need to know how to find the information".    That was some of the best advice I've ever gotten.