Skip to content

Author: r.kyser

Published on Leave a comment on Quackery: A Comic

Hello all. For today’s blog post, we’re sharing a comic from one of our reference librarians, regarding quackery. Enjoy! 

Quackery: An Introduction to Fakes, Frauds, Forgeries and Other Falsehoods.
Image: A librarian, Rebecca, stands in the middle of the panel, one finger raised. She is a woman with pale skin, curly brown hair tied back in a ponytail, a pair of glasses, and she wears a purple short sleeved polo shirt. Narration: Quackery: A term that refers to the practices of fraudsters and conmen, often in the field of medicine. Rebecca, Dialogue: “And let me tell you, history has a lot of quacks!”
Image: A pop up is shown, with bright outlandish colors. The ad states “You Won! Miracle Acne Cream. Buy now!! $19.99.” A mouse icon hovers over the “buy now” text. Narration: Folks have tried to scam each other…
Image: A man with light brown skin and short brown hair wears a toga and stands in front of a sales booth. He is yelling, gesturing to his wares. The sales booth has some coins and a bowl on it, and on the front, a sign says “100% effective, no refunds” Narration: Probably as long as lying was invented. Greek Salesman Dialogue: “Plague B-Gone charms! One silver a piece!”
Image: Rebecca shrugs her shoulders, rolling her eyes, as she stands in the center of the panel. Rebecca Dialogue: “There tends to be some trends when it comes to health fraud and fakery.”
Image: A radioactive warning sign is in the center of the panel, surrounded by four objects, text and a small version of Rebecca. The text states “We put radium in everything!” The following products are shown, each labeled: toothpaste, hair cream, water (with the text “oh god why don’t” on the label”) and pottery (a small goblet). The chibi Rebecca says “I mean everything.” Narration: Trend 1: A new substance is discovered or an old one becomes trendy and someone decides to sell it for its “health benefits.” (1)
Image: Three different examples are shown. The first is the drawing of a class of water with the label “raw water” and the description: “free dysentery in every bottle.” The second is a drawing of a vaccine shot with the label “anti-vax solutions” and the description “Diphtheria: the comeback kid!” The third is a small picture of the sun wearing sunglasses with the label “types of sun bathing” and the description “skin cancer? Never heard of her.” The drawing of the sun says “aw no.” Narration: “Trend 2: Folks have been arguing that “All Natural” solutions are the golden ticket- for a pretty penny of course.
Image: A light skinned homeopath with a vest and a beard stands in front of a table with a snake and a bottle on it. The homeopath’s eyes are closed, and he speaks to the snake as if lecturing it. The snake replies. Narration: Trend 3: Some strands of quackery just won’t die even when they fail to work. Homeopathy for example, keeps on kicking. (2) Dialogue, Homeopath: “If I dilute your venom dozens of times, I can treat-” Dialogue, Snake: “Abssssoultely nothing.”
Image: Rebecca stands in the middle of the panel, looking serious. Rebecca Dialogue: “As silly as some of these remedies sound, quackery is dangerous.”
Image: Small cartoon drawings of measle viruses fill the panel. Narration: Both on an individual level; patients can die from preventable/treatable conditions. (Measles, for example, has made a big come-back in the US with the rise of the Anti-Vax movement). (3)
Image: A diagram of a cartoon human is in the center of the panel, with lines dividing their brain into different color coded sections with numbers. On the right, there is text identifying a trait to correspond with each part. It reads as follows: “1. Prophecy 2. Intimidation. 3. Immortality 4. Mirth. 100% B.S) Narration: But also at a societal level where it can uphold harmful ideas. Phrenology for example was used to uphold and codify scientific racism.
Narration: People have combatted quackery for ages and it’s easy to join their ranks. Here’s some ways to help. Image Description: A duck wearing a lab coat floats in a body of water, Dialogue, off screen: “...I don’t think that duck has a medical degree.”
Narration: 1. Be skeptical! Anyone can fall for quackery regardless of intelligence. Never assume you’re “too smart” to fall for fraud. Image: A picture of Garfield on a computer screen appears. It is just his face looking straight forward. A cursor is a few centimeters away. Under Garfield text states as follows: “you are not immune to propaganda) Narration: 2. Warn friends and family about fake treatments. Don’t shame people but be honest and clear about your concerns. Image: A women with dark hair in moon buns and brown skin speaks to a man with light skin and freckles. She looks concerned, and he looks confused, holding a tablet in his hands. Behind them is living room furniture.
Narration: 3. The modern medical system has its problems but if anyone tries selling you something that “doctors are hiding from you” it's probably quackery. Image: Rebecca wearing the same outfit as Fred from scooby doo, holds a mask in her hands. She faces a human sized duck, who is tied up and glaring at her, wearing the rest of the costume to which the mask belongs. Rebecca Dialogue: “Just as I suspected…”
Narration: 4. If you see an obvious scam online, report it! Or draw it into your comic on medical scams. In parentheses under the box, text reads as follows: “no really, this is based on a real tiktok” Image: A phone is pictured showing tiktok. On the screen is a woman with light skin and red hair hugging a large oak tree. The text says “tree vibrations for health” at the top and “health secrets at the bottom”

Citations:

  1. Kang L, Pedersen N. Quackery : A Brief History of the Worst Ways to Cure Everything. Workman Publishing; 2017.
  2. Dotinga R. The lethal legacy of early 20th-century radiation quackery. Washington Post. https://www.washingtonpost.com/health/the-lethal-legacy-of-early-20th-century-radiation-quackery/2020/02/14/ed1fd724-37c9-11ea-bf30-ad313e4ec754_story.html. Published February 14, 2020. Accessed March 29, 2024.
  3. The worrying return of a nasty virus: Better awareness and access are needed to rebuild measles vaccination rates. Financial Times. January 29, 2024:22.
  4. Rogers K. Phrenology | History, Theory, & Pseudoscience | Britannica. Encyclopaedia Britannica. Published March 14, 2024. Accessed April 8, 2024. https://www.britannica.com/topic/phrenology

Published on

For a moment, let’s entertain a hypothetical. Let’s say you have an excellent paper on your hands about the impact of smoke on the lungs. Your team is about to submit it for publication: pretty exciting! When you get your paper back from the publisher, it’s mostly good news: they’re willing to publish your paper with the caveat that you add a diagram of the lungs to your paper as a visual aid of the systems impacted. The problem? You have no idea where you could acquire an image that would suit this task that wouldn’t potentially violate copyright. 

With this conundrum, one of your coauthors suggests a solution: why not generate one? They have a subscription to Midjourney, the AI software that can generate images from text. Why not give Midjourney a summary of the diagram you need, have it generate it, and then use that for your paper. After checking the journal’s policies on AI (it’s allowed with disclosure), you do just that, glad to have quickly moved past that stumbling block. 

Pretty great, right? It sure sounds like it, until you take a look at the images Midjourney generated. Because on closer inspection, there are some problems. 

Below is an image I generated in CoPilot for this blog post. I didn’t ask it to do something as complicated as making a diagram of how smoking impacts the lungs; instead I asked for just a diagram of human lungs. Here is what I got, with my notes attached.

An image of an AI generated diagram of the lungs in a human women is featured with red text boxes pointing to errors. In the upper left, a box says "nonsense of gibberish text" and a red line points to oddly generated letters that mean nothing. Below it, another box reads "I don't know what this is supposed to be, but I don't think it's in the armpit" with a line pointing to what looks to be an organ with a flower pattern in it. Below that, another box reads "this heart is way too small for an adult" and the red line points to the heart on the diagram. On the left, the top red box reads "now the stomach does not reside in one's hair or arteries" with red lines pointing to a picture of the stomach that is falsely labeled as being in the hair or neck. Below that, a new box reads "what are the gold lines supposed to be in this diagram" and it points to yellow veins that run through the figure like the red and blue that usually denote the circulatory system. The last box on the right says "I have no idea what this is supposed to be" and points to what looks to be bone wrapped around a tube leading out of the bottom of the lungs.

Alright, so this might not be our best image. Thankfully, we have others. Let’s take a look at another image from the same prompt and see if it does a better job. 

An image of an AI generated diagram of the lungs is featured with red text boxes pointing to errors. In the upper left, a box says "more nonsense text" and a red line points to oddly generated letters that mean nothing. On the right side, a box says "bubbles should not be in the lungs!" with a red line pointing to what looks to be odd bubbles inside the lungs. Below it, a red box reads "what are these small clumps/objects?" and it points to what looks to be red large bacteria and clumps on the lungs.

So what happened here? To explain how this image went terribly wrong, it’s best to start with an explanation of how AI actually works.

When we think of AI, we generally think of movies like The Terminator or The Matrix, where robots can fully think and make decisions, just like a human can. As cool (or terrifying depending on your point of view) as that is, such highly developed forms of artificial intelligence still solely exist in the realm of science fiction. What we call AI now is something known as generative AI. To vastly simplify the process, generative AI works as follows: you take a computer and feed it a large amount of information that resembles what you want it to possibly generate. This is known as “training data.” The AI then attempts to replicate images based on the original training data. (Vox made a video explaining this process much better than I can). So for example, if I feed an AI picture of cats, over time, it identifies aspects of cats across photos: fur, four legs, a tail, a nose,etc. After a period of time, it then generates images based on those qualities. And that’s how we get websites like “These Cats Do Not Exist.”

If you take a look at “These Cats Do Not Exist” you might notice something interesting: the quality of fake cat photos varies widely. Some of the cats it generates look like perfectly normal cats. Others appear slightly off; they might have odd proportions or too many paws. And a whole other contingent appears as what can best be described as eldritch monstrosities.  

The reason for the errors in both our above images and our fake cats is due to the fact that the AI doesn’t understand what we are asking it to make. The bot has no concept of lungs as an organ, or cats as a creature; it merely recognizes aspects and characteristics of those concepts. This is why AI art and AI images can look impressive on the surface but fall apart under any scrutiny: the robot can mimic patterns well enough, but the details are much harder to replicate, especially when details vary so much between images. For example, consider these diagrams of human cells I had AI generate for this blog post.

A picture of an AI generated human cell. There are red boxes with text pointing out errors and issues in the image. The top box has the text "nonsense words. Some of these labels don't even point to anything" with two red lines pointing to a series of odd generated letters that mean nothing. Below that, a red box has the text "I have no idea what this is supposed to be" with a red line pointing to a round red ball. On the right side, a text box reads "is this supposed to be a mitochondria? Or is it loose pasta?" with a red line pointing to what looks to be a green penne noodle in the cell. Below that, a red text box reads "I don't think you can find a minature man inside the human cell" and a red line points to the upper torso and head of a man coming out of the cell.

Our AI doesn’t do bad in some regards: it understands the importance of a nucleus, and that a human cell should be round. This is pretty consistent across the images I had it make. But when it comes to showcasing other parts of the cell we run into trouble, given how differently such things are presented in other diagrams. The shape that one artist might decide to use for anaspect of a cell, another artist might draw entirely differently. The AI doesn’t understand the concept of a human cell, it is merely replicating images it’s been fed. 

These errors can lead to embarrassing consequences. In March, a paper went viral for all the wrong reasons; the AI images the writers used had many of the flaws listed above, along with a picture of a mouse that was rather absurd. While the writers disclosed the use of AI, the fact these images passed peer review with nonsense text and other flaws, turned into a massive scandal. The paper was later retracted. 

Let’s go back to our hypothetical. If you need images for your paper or project, instead of using AI, why not use some of Himmelfarb’s resources? On this Image Resources LibGuide, you will find multiple places to find reputable images, with clear copyright permissions. There are plenty of options from which to work. 

As for our AI image generators? If you want to generate photos of cats, go ahead! But leave the scientific charts and images for humans. 

Sources:

  1. Ai Art, explained. YouTube. June 1, 2022. Accessed April 19, 2024. https://www.youtube.com/watch?v=SVcsDDABEkM.
  2. Wong C. AI-generated images and video are here: how could they shape research? Nature (London). Published online 2024.
Viewing Message: 1 of 1.
 Notice

This site uses cookies to offer you a better browsing experience. Visit GW’s Website Privacy Notice to learn more about how GW uses cookies.