Google is pushing hard to get us to use AI for all kinds of things. We should just ask Google our questions about biology!
After all, I, as a biologist have so much confidence in the power of AI to address difficult questions about biology. For instance, ask it to explain an ovarian cyst to you.
(I’ve put this image below the fold to avoid triggering nightmares or confusion.)
Lovecraft had nothin’ on AI. Stare deep and feel the madness rise within you.
Perhaps it can handle gynecological and embryonic anatomy a little better? I dare you to contemplate the outcome of this pregnancy.
I’m happy to report that we human biologists still have some job security. And Google/ChatGPT/Grok can go crawl up their own rectum, if they can find it.
The Garden of Earthly Delights? Anyway, what is the name of that film director that has been into body horror since the 1980s? He can now use LLMs to come up with new ideas.
@Birgerjohansson, Frank Henonlotter? There’s more than one.
Last I checked I don’t poop out of that hole, but at least we have a visual for how the pregnancy in the movie Junior works :P
Gotta say both the tiny hands as ovaries and the pregnancy image gave me a good laugh this morning :D
@Birgerjohansson, # 1: Clive Barker. The story is “Midnight Meat Train.”
Never train your AI with the work of Charles Burns.
Oops-Clive Barker is the author, not the director. The director is Ryuhei Kitamura.
You’ll have to admit that the hallucinations can be amusing. This does illustrate the point: never rely on a single source for information, particularly if the source is still in beta (or perhaps alpha) test. And if using AI, adjust your prompt.
This is the response to a query to DuckDuckGo’s “Search Assist” I had a few months ago: What is complete knee replacement?
That’s not wrong and it has links to Wikipedia, an OK source for info, and the Cleveland Clinic, a relatively reliable source of info on medical conditions.
The extended “More” section provides more detail although under “Recovery and Outcomes: Post-Surgery Care” in says, “Rehabilitation: Physical therapy is crucial for regaining strength and mobility. Patients are encouraged to start moving soon after surgery.” That doesn’t go far enough to characterize the rehabilitation process which is hours of PT often painful nor the prospect of a further procedure called a “Manipulation Under Anesthesia.”
AI is just more of the insanity, waste and abuse of the tech ahole billionaires:
G00GLE: has always stolen peoples creative works. It is spyware compiling massive data on everyone
Imbeciles using AI instead of regular searching are unaware of the horrible price our environment and society are paying
one example of many:
https://kanoppi.co/search-engines-vs-ai-energy-consumption-compared/
energy use – AI 10X higher per query
pollution & emissions – AI 340x higher per query
daily energy use – AI 58x higher
And, AI using fools are getting bullshit answers most of the time
Just got back from an appointment with my nephrologist. Kidneys are doing mostly fine. But apparently the labs found that my Parathyroid glands might be overproducing. My dad did a quick Google search on high PTH and asked if it was linked to weight gain, since that’s an ongoing concern they have about me. Search said yes, but then we saw that was Gemini talking. Fortunately we had the actual doctor there to tell us otherwise. No, thank you, Google. I’ve been using Duck Duck Go with “noai” near the start of the URL and its AI settings as off as I can get them.
@christoph Clive Barker may not have directed Midnight Meat Train, but he did direct Night Breed (and the original Hellraiser) which has elements of body horror and also features the director @BirgerJohansson was probably actually thinking of: David Cronenberg
Taking the “I” out of AI!
Circulating that last image will greatly reduce fellatio stats around the world, at least among the literate.
And, here is another interesting viewpoint on AI:
https://featureassets.gocomics.com/assets/3160ab905e8d013eaf7f005056a9545d
We NEVER use G00GLE. For those smart enough to use duckduckgo, they provide an option to dump AI:
https://duckduckgo.com/duckduckgo-help-pages/ai-features/opting-out-of-ai
It’s funny because this fits the performative anti-AI hyperventilating to a T.
The cat’s out of the bag. People are using machine learning because it’s a good tool. People can misuse good tools. Even if you plug your ears and wish all machine learning goes away, people will still use it. Your subculture only looks more and more ridiculous as you dig in your heels.
All chatbots are AI, but not all AI is chatbots. People are clearly still confused by that.
As for its lack of utility: https://phys.org/news/2025-08-ai-biological-variability-high-serum.html
@14 beholder wrote: Even if you plug your ears and wish all machine learning goes away, people will still use it. Your subculture only looks more and more ridiculous as you dig in your heels.
I reply: I’m not sure what you are getting at. No one here is a Luddite (including me) and no one suggested that AI go away. There are appropriate uses of AI. But, to use it all the time is irresponsible and wasteful. The intelligent person does not use a sledgehammer to push in every thumbtack.
Also, @15 John Morales makes an informative point. Even though, I personally, I’m not interested in chatting with a disembodied entity.
This came out today.
95% of organizations that invested in AI are getting a zero return on that investment.
The article goes on to point out that improvements to productivity from new technologies can take many years or decades to happen.
This was true for computers, biotechnology, the internet, cell phones, genomics, etc..
It takes time for a new technology to spread out and displace or augment existing ways of doing things.
It could take a decade or two for AI to actually show what it can do.
I’ve already dealt with AI agents a few times and they completely failed.
I sent an email to Moderna asking when their latest updated mRNA vaccine will be released. This usually happens in the late summer or early fall for seasonal viruses.
I got a long reply about where to report my vaccine induced injury, which had nothing to do with my question.
Which was followed by a followup email that said since I didn’t file a report on my nonexistent vaccine injury, my case was discontinued.
@ killyosaur, #9, and to everyone else here talking about body horror
I would have guessed David Cronenberg as well although, to be pedantic, he actually got his start with body horror in the 1970s with movies like Shivers (1975) and Rabid (1977), two extremely warped takes on zombies and vampirism.
In the former, a mad scientist develops sexually-transmitted parasites that turn the infected into sex-crazed not quite zombies (they can still think and reason to some degree, but only in service of spreading the infection). The scientist wanted to reduce the world to “one big beautiful orgy.” The latter involves an experimental medical technique that turns a badly-burned woman (played by Marilyn Chambers) into a kind of vampire, whose sting turns people into more traditional zombies.
A couple of other Cronenberg films that deal fairly extensively with body horror are The Brood, Videodrome, and The Fly (the best, in my opinion).
Fun fact (which you may already be aware of): the serial-killing psychiatrist in Nightbreed is played by none other than David Cronenberg.
Nightbreed, based on Barker’s Novel Cabal, is actually the movie that nearly drove Clive Barker out of Hollywood; the movie is a mess, and it’s all thanks to executive meddling. The suits didn’t understand that the “monsters” of Midian are not the bad guys. They mostly just want to be left alone- the real monsters are the human beings, represented as they are by aforementioned serial killer, a brutally-fascist police chief (or maybe sheriff, I don’t remember), and the rednecks and religious fanatics who want to destroy anything different from themselves.
The suits insisted on hacking up the movie to make it look like the freaky-looking creatures were the real monsters, which, being the opposite of what the movie was about, resulted in an incoherent mess. Barker was rightfully outraged, and nearly swore off movie-making for good (he returned to directing one last time in 1995 with Lord of Illusions). His director’s cut restores some semblance of the original intent, but its still missing of a lot the movie as it was meant to be. Personally, I recommend reading the book before seeing the director’s cut, to better understand what it’s meant to be (the comics and the short story collection Midian Unmade are also worth a look, if you can find them).
Clive Barker also directed Hellraiser, based on his novel The Hellbound Heart, both of which I recommend to anyone who can stomach extreme gore and body horror. Actually, I recommend all of Barker’s work to those who don’t mind the grue; his reputation as part of the “splatterpunk” generation of horror writers belies the depth and complexity of his writing. Though his work is often quite dark, it also carries with it a sense of wonder and appreciation for stranger forms of beauty.
Quite a few of Barker’s written works have been translated (often poorly, it must be said) into film, including a number from his excellent short stories collections The Books of Blood. Midnight Meat Train has already been mentioned from that collection, but there’s also Rawhead Rex (one of my favorites from the collection, but the film completely misses the point of a lot of the religious/sexual/gender symbolism of the story). My other two favorite stories, “The Skins of the Fathers” and “In the Hills, the Cities,” would be difficult to adapt to film well. He’s also written quite a few screenplays, like Candyman and Underworld.
I used ChatGPT a few years ago to ask about a book I was reading, The Daughter of Doctor Moreau by Silvia Moreno-Garcia, and it started babbling on about The Madman’s Daughter by Megan Shepherd. Oops.
Recently I’ve been querying Gemini and Perplexity about William by Mason Coile, which is ironically a horror story about an AI who loves Mephistopheles in Faust as a role model and takes control of then haunts a smart home. The AIs I interact with are surely playing close to stuff people have said on various websites/blogs about the book, but they aren’t straying too far from my understanding of the book and have enhanced the reading experience. I am currently in a triad between William, Faust, and Frankenstein and ironically am having AI help me out with this interest, given the malevolent AI in William. I throw stuff at the AI wall and analyze the output with discernment. I can separate wheat from chaff. I am also engaging in non-AI content.
I think the AI generated anatomy mishaps PZ focuses on are low hanging fruit. I have asked Canva’s AI to generate image output for flyers several times. There are elements you just can’t find in the stock images. You have to sift through some nonsense (slop) for the good stuff. There were unintended masterpieces, like three eared rabbits or rabbits shooting volcano lava out their ass (while surfing?) I just had to keep. But I was asking Canva to do ridiculous stuff and it obliged. Details like fingers or wording on marquee signs are epic fails.
AI can be fun, and Gemini or Perplexity are sometimes not too bad depending on the query. I’m not relying on them for legal or medical advice.
I should maybe try ChatGPT again sometime.
This kind of “AI” stochastic parrot is just another tedious bit of capitalist hype, nothing more. A quirky little toy that can be mildly amusing to play with, and a boon to plagiarists and lazy people who expect they won’t get caught, but ultimately of little value. I look forward to its eventual demise in a few years’ time.
I have read numerous cheap summaries of books over the years and some of that stuff is allegedly written by humans and who have no idea what they are talking about or at least get key details wrong. I wonder if some of it was generated by an LLM and uncritically published for cash. IMO SparkNotes and SuperSummary are usually pretty good or at least rise above the rest for actual human generated content.
Youtube book content varies too. Some of it is obvious AI slop. Some human generated content is not much better than AI. There are some awesome videos I have found on Youtube for book analysis but these are book dependent. Anything that seems AI based is out the door quickly.
I have recently dabbled in AI to see how it engages book content. I find the challenge in analyzing the output is like when I’m engaging a person who may have read the book and I can determine how askew their take might be in comparison to my own. That works for me. Plus I am intentionally engaging with AI and not encountering bought book summaries or watched Youtube channels that are clandestine AI BS. Big difference.
@18 truth be told, I mentioned Henonlotter first. And he got his official film start in 82 with Basket Case. So that would be the more accurate one to bring up :D
Also. my personal favorite Clive Barker novel was “The Great And Secret Show ” which features a “guest appearance” as it were from the main character of Lord of Illusions :P
Is it being too Cap’n Obvs to note whatever you use to search you should try to find multiple preferably independent sources to confirm what is claimed?
But yeah, that AI shit, fucking hell.. WTF.
“Is it being too Cap’n Obvs to note whatever you use to search you should try to find multiple preferably independent sources to confirm what is claimed?”
You’re missing the point.
You ask the AI to try to find multiple preferably independent sources to confirm what is claimed.
—
See, another salient aspect is that one should know how to actually specify what one seeks. One of many.
Put it this way: what is there to stop you from storing the text string ”and whatever you use to search you should try to find multiple preferably independent sources to confirm what is claimed” and then appending it to each query?
;)
They’ve really got to stop training AIs on Cronenberg movies
My primary exposure to AI has been as an assist in writing Python programs. The technology is getting better. I used to get frustrated when the generated code called Python libraries that either didn’t exist or used the wrong arguments, but that has been less of a problem lately. As I see it AI (specifically Large Language Models, or LLMs) can be very useful, but problems arise when people think it is magic (or sentient) and treat it as such. It will be a while before it becomes something that can be fully trusted, but I look forward to it becoming more capable, reliable, and less of a power hog.
Given that half the doctors in the country graduated in the bottom 50% of their class, I’ll be glad when AI can provide a baseline competency for things like medical support without the biases or lack of curiosity my wife and I have encountered from doctors over the years. Also, as I get older I look forward to having a digital assistant that can help me keep track of things. We’re still in the toddler stage of this technology, and people with that AI hammer are still trying to figure out what constitutes a valid nail.
Here is how competent microsoft and AI are:
https://www.ghacks.net/2025/08/26/copilot-launches-in-excel-but-microsoft-warns-against-using-it-for-any-task-requiring-accuracy/
And here is now helpful AI is:
https://azmirror.com/2025/08/26/repub/ai-is-making-it-easier-for-bad-actors-to-create-biosecurity-threats/
ROFLMAO at AI and any idiot that thinks it is competent