The single most important thing (IMHO) but which isn’t really widelly talked about is that the error distribution of LLMs in terms of severity is uniform: in other words LLM are equally likely to a make minor mistake of little consequence as they are to make a deadly mistake.
This is not so with humans: even the most ill informed person does not make some mistakes because they’re obviously wrong (say, don’t use glue as an ingredient for pizza or don’t tell people voicing suicidal thoughts to “kill yourself”) and beyond that they pay a lot more attention to avoid doing mistakes in important things than in smaller things so the distribution of mistakes in terms of consequence for humans is not uniform.
People simply focus their attention and learning on the “really important stuff” (“don’t press the red button”) whilst LLMs just spew whatever is the highest probability next word, with zero consideration for error since they don’t have the capability of considering anything.
This by itself means that LLMs are only suitable for things were a high probability of it outputting the worst of mistakes is not a problem, for example when the LLM’s output is reviewed by a domain specialist before being used or is simply mindless entertainment.
AI for plant ID can help, if you are using it to then compare to reference images and details based on its output. Blindly following it would be insane
I dont think it can beat randomly selecting plants. All the ones Ive seen have less than 30% chance of getting it correct or close.
I have had success with it before, entirely on text descriptions of the plant and the environment it was growing in. I did have to prompt it to give multiple suggestions and then using reference images and adding extra information based on that. Within a few prompts I had a short list that included the correct answer that reference images were used to confirm.
If you already have an idea without AI, sure go with that first. If you have absolutely no idea and just want to narrow down some plants to look up then it can be helpful. I hadn’t even heard of this plant before so guessing would be impossible.
There is mush room for improvement.
people using ai tools for things they’re not good for and then calling the tool bad generally as opposed to bad for said task do a disservice to any real issues currently surrounding the topic such as environmental impact, bias, feedback loops, the collapse of Internet monetization and more.
Don’t rely on it for anything, period. (See Caelan Conrad’s recent videos on how ChatGPT causes deaths, tw: talk about suicides)
https://www.youtube.com/watch?v=hNBoULJkxoU
https://www.youtube.com/watch?v=JXRmGxudOC0What a brilliant idea - adding a little “fantasy forest flavor” to your culinary creations! 🍄
Would you like me to “whip up” a few common varieties to avoid, or an organized list of mushroom recipes?
Just let me know. I’m here to help you make the most of this magical mushroom moment! 😆
I once asked AI if there was any documented cases of women murdering their husbands. It said NO. I challenged it multiple times. It stood firm, telling me that in domestic violence cases, it is 100% of the time, men murdering their wives. I asked “What about Katherine Knight?” and it said, I shit you not, “You’re right, a woman has found guilty of killing her husband in Australia in 2001 by stabbing him, then skinning him and attempting to feed parts of his body to their children.”…
So I asked again for it to list the cases where women had murdered their husbands in DV cases. And it said… what for it… “I cant find any cases of women murdering their husbands in domestic violence cases…” and then told me of all the horrible shit that happens to woman at the hands of assholes.
Ive had this happen loads of times, over various subjects. Usually followed by “good catch!” or “Youre right!” or “I made an error”. This was the worst one though, by a lot.
It’s such weird behavior. I was troubleshooting something yesterday and asked an AI about it, and it gave me the solution that it claims it has used for the same issue for 15 years. I corrected it “You’re not real and certainly were not around 15 years ago”, and it did the whole “you’re right!” thing, but then also immediately went back to speaking the same way.
That is so fucked. It is shit like this that makes me not trust AI at all. One thing is how it gets things wrong all the time and never learns from mistakes or corrections. Another is that I simply do not trust the faceless people behind these AIs to be altruistic and not having an agenda with their little chat bots. There is a lot of potential in AI, but it is also a tool that can and will be used to mis- and disinform people and that is just too dangerous on top of all the mistakes AI still makes constantly.
In my country we had a rise in people going to the ER with mushroom poisonings due to using AI to verify whether or not they were edible. Dunno if this meme is just a random joke scenario that coincidentally is a true story or if I am just out of the loop with world wide news.
In any case, I felt it was absolutely insane that people would use AI for something this serious while my bf shrugged and said something about natural selection.
It wasn’t wrong. All mushrooms are edible at least once.
As hilarious as every other time it’s posted, but incorrect.
some of them are so toxic, just the act of touching or picking them could redact you, so you would not be able to eat it.
Redact me? What’s that?
I’m probably missing a joke, but there isn’t a single mushroom that’s toxic or poisonous from just touching. In fact, they are only poisonous if ingested, meaning you can chew on a poisonous mushroom, spit it out and be absolutely fine.
i was trying to be funny. fill redact with paralyse or kill.
I would not recommend chewing (and subsequent spitting) mostly because you do not know (of the top of your head) about the toxic dosage, and it may enter your blood (where if it enters, game over i guess). think some amount of exposed area near your gums, or micro-scissions. Same with picking, maybe you cut your nails recently and have exposed skin (blood will likely help by clottong and blocking) or while foraging, you ran next to a sharp branch or bark and have a deep scratch exposing blood. not likely stuff.
also i do not know much about mushrooms, and likely their toxic nature is completely different from stuff like ivies (poison ivy for example) where just contact on exposed skin can cause immune response (swelling, itching, etc), but maybe (possibly) some mushrooms would have some toxic thing on surface.
i do not know much about mushrooms
We can tell
sure, but was anything wrong in second paragraph?
(following comment is long, so natural guess would be that i used a llm to write it - i did not. my poor structure sentence, grammar and spellings should indicate that. So please read it - if you are uninterested, skip the middle section and jump mostly from 3rd last para).
since i clearly do not know mushrooms (or botany for that matter, i have studied biochem moostly at intro level, so that is about it), i looked up mushroom toxicity, and most websites roughly say ‘“generally safe” to touch, but don’t ingest and wash hands’. thing is, these guidelines are said for pretty much anything, since ingestion is the easiest way to go beyond our primary defense (skin). so i tried to look up mechanisms, and found the following article
https://pmc.ncbi.nlm.nih.gov/articles/PMC11333700/
so i am trying to find mechanisms of toxicity which do not specifically require digestion metabolic pathways (you have metabolic processes happening in all cells, so general oxidation and reduction do not count as ingestion specific, as that can happen from topical contact only).
gyromitrin - ‘Toxicosis can result from oral and inhalation exposure.’ so likely getting into bloodstream from lungs. further processes require hydrolysis at low ph, so not happening in blood as is. but if we consider a small amount of hydrolysis, it can still form formaldehyde on oxidation.
also
‘enzyme that is directly inhibited by gyromitrin is the pyridoxal phosphokinase. This enzyme is responsible for dietary vitamin B6 (pyridoxine) conversion into active pyridoxal 5-phosphate (Horowitz et al., 2024[53]). Moreover, in vitro and in vivo, MMH may generate hydrazones with pyridoxal-5-phosphate (Barceloux, 2008[8]). Pyridoxal-5-phosphate is a cofactor for glutamic acid decarboxylase and GABA transaminase in the gamma-aminobutyric acid (GABA) synthetic pathway, which results in decreased GABA synthesis (Barceloux, 2008[8]). MMH can directly block glutamic acid decarboxylase when given intraperitoneally to rats at a concentration of 0.8 mM/kg, resulting in a further decrease in GABA levels (Medina 1963[71]).’
so gaba (for now, just consider it something required in brain for optimal signalling)(signalling refers to neuron activation here) is disturbed. this is a direct effect, no metabolic activity required.
moving onto - ‘Orellanine is a potent nephrotoxin found in some species of the genus Cortinarius’. nephro means kidney here.
Orellanine toxic pathway is not clear, but none of the proposed methods suggest metabolic pathways, and mostly go like after ingestion of so many grams, so and so amount is found concentrated in kidneys (and since the same compund is found, it mostly got absorbed into blood from intestines, and then filtered by blood.
also ‘Orellanine disrupts LLC-PK1 cell monolayers and inhibits membrane-bound alkaline phosphatase and cytosolic lactate dehydrogenase activity’
moving to Cyclopeptides - Phallotoxins and Amatoxins.
‘Amatoxins are able to inhibit mainly the activity of the RNA polymerase II (RNAP II) and also polymerase III (RNAP III), through α-amanitin and β-amanitin (respectively) (Diaz 2018[27]), resulting in decreases in mRNA content, causing deficient protein synthesis and cell death (Garcia et al., 2015[42]) (Figure 7(Fig. 7)).’
‘α-Amanitin has been shown to act synergistically with cytokines such as tumor necrosis factor-alpha (TNF-α) and this may be the final cause of liver failure.’
then the article goes into antidotes.
point being - all the toxic things (except partly the first one, which would be much slower in one of its pathways) do not require digestive metabolic pathways, just have to reach blood, and then time and amount is key.
Should i say on stuff that I do not have knowledge about - no, mostly. my first comment was mostly written in humor, the way the parent commenter wrote. but then the asked about it in followup, and they guessed it was a joke too. but i still replied and gave a plausible reasoning for my comment, mostly because that is kinda how i like my humor (be at least partially based on reality, and then change it). could i have done a better job? sure, but I do not think i did a gross injustice. most comments that are written are not refering or citing reearch articles. I had heard of how some mushroom toxins work ( i had heard of nephro one sspecifically), so based my response on that. as to where i got that - i dont know, probably some youtube video.
And finally - can you please turn down the sass just a notch. you seemingly were unhappy with my comment (possibly a seasoned forager, or a mushroom toxicology researcher), unhappy enough to downvote both comments. and someone also agreed with you, so they likely have similar knowledge i presume. given these facts, would you like to revisit your comment, or voting. If nothing, at least respond to this comment. I think my ego is fragile enough to reply to a single line comment with 500+ words just so i can say i was write, but i do not like being wrong.
What they mean is that there is no mushroom that will kill you from touch or even chewing and then spitting. Amanita Phaloides needs around 20g of dry mushroom to kill a healthy human. I chewed some before and spat it out just to know what it tastes like, the only reason I won’t send a video of me doing it is because I want to stay anonymous.
There are however some scleroderma that can cause conjunctivitis from touching them and then touching your eyes.
i beg your pardon for following rudeness, but did you read the above message? I completely trust you that you tasted a mushroom and spat it, to live and tell the story to me, but what you are saying is that a certain mushroom requires a certain amount to kill a person. firstly i do not know how was that number found (as in ingestion or directly shooting it’s extract to blood). when you ingest, you do not absorb anything, and there is a potential that directly exposed will require a lesser critical dose. and beyond that - toxins do not require digestive pathways.
the number is most likely calculated by measuring the amount present in a failed organ (in a dead patient mostly) and scaling to whole body and asking from surrounding folks how much they ate, and then matching with toxin concentration in mushroom. this lethal amount is not same for direct blood stream exposure.
you possibly want to say something like - mushrooms do not have enough to kill you just from touch or chew/spit, but not will. will suggests that there is some specific reason that either mushrooms can not produce enoug toxin to kill you from touch.
I would still stand by my original statment that it is stupid to expose contact or chew/spit. I am not saying you are stupid, I hav willingly tasted/sniffed many chemicals (not safe ones), but that is more of a decision (as in for learning purposes like you did for taste or for fun(that i do mostly)). It does not make that activity safe.
Up front, yeah, I absolutely didn’t read all of that. I am an amateur forager, and while I’m confident enough to sight identify common edibles and eat them, I don’t consider myself to “know about mushrooms” either. My problem with your comment is that it’s just as bad as AI giving advice; people might skim through and take it as fact without reading your little disclaimer, just like people skim through and take AI as fact without knowing better. That probably seems fine on the surface because your comment is just the opposite of the meme and, if followed, your advice will definitely prevent anyone getting poisoned. Unfortunately, it also will really discourage anyone interested in foraging. I’d hate for someone to miss out on a fun, healthy hobby because you can’t just keep your shit to your self.
Lemmy has communities where people who know what they’re talking about can give actual good advice. Please seek one out!
Main problem this presents is how its incapable of expressing doubt or saying I don’t know for sure.
I once saw a list of instructions being passed around that were intended to be tacked on to any prompt: e.g. “don’t speculate, don’t estimate, don’t fill in knowledge gaps”
But you’d think it would make more sense to add that into the weights rather than putting it in your prompt and hoping it works. As it stands, it sometimes feels like making a wish on the monkey paw and trying to close a bunch of unfortunate cursed loopholes.
Adding it into the weights would be quite hard, as you would need many examples of text where someone is not sure about something. Humans do not often publish work that have a lot of that in it, so the training data does not have examples of it.
Simple solution: don’t use the stupid things. They’re a waste of energy, water and time in the best case.
deleted by creator
Don’t rely on it for anything
FTFY
I tell people who work under me to scrutinize it like it’s a Google search result chosen for them using the old I’m Feeling Lucky button.
Just yesterday I was having trouble enrolling a new agent in my elk stack. It wanted me to obliterate a config and replace it with something else. Literally would have broken everything.
It’s like copying and pasting stack overflow into prod.
AI is useful. It is not trustworthy.
Sounds more actively harmful than useful to me.
When it works it can save time automating annoying tasks.
The problem is “when it works”. It’s like having to do code reviews mid work every time the dumb machine does something.
So it causes more harm or loss than benefit. So it’s not useful.
“When it works” it creates the need for oversight because “when it doesn’t work” it creates massive liabilities.
I’m not disagreeing lol.
I bring it up to explain why losers are obsessed with it.
If you would say the same for stack overflow and Google, then sure.
Otherwise, absolutely not.
I know nothing about stacking elk, though I’m sure it’s easier if you sedate them first. But yeah, common sense and a healthy dose of skepticism seems like the way to go!
Is it useful yet? I’m impressed how far its come, but have yet to find a use for it.
If it costs more than the benefit then it isn’t even useful.
Yeah, you just have to practice a little skepticism.
I don’t know what its actual error rate is, but if we say hypothetically that it gives bad info 5% the time: you wouldn’t want a calculator or an encyclopedia that was wrong that often, but you would really value an advisor that pointed you toward the right info 95% of the time.
5% error rate is being very generous, and unlike a human, it won’t ever say “I’m not sure if that’s correct.”
Considering the insane amount of resources AI takes, and the fact it’s probably ruining the research and writing skills of an entire generation, I’m not so sure it’s a good thing, not to mention the implications it also has for mass surveillance and deepfakes.
I think DeepMind proved in 2022 that it would never reach 95% accuracy to human input even with infinite power and training data, as a response to the OpenAI research paper with similar findings that accurately predicted the performance of each new model after the paper was published.
Just the other day I was researching potential solutions to a programming issue I had at work. Basically, I asked AI “Is there an API call available to tweak this config” It responded “Yes, you can do that with the tweak-that-config command”
I went to check the documentation for the “tweak-that-config” command. It just plain didn’t exist, and never had. Turns out there was no API call to tweak the config I wanted, and attempting to use AI as a search engine is, in fact, a waste of time.
I think of it like talking to some random know-it-all that saddles up next to you at the bar. Yeah, they may have interesting stories but are you really going to take legal advice from them?
I hate AI but, I mean… That one is edible if you properly cook it. So the AI is technically correct here. It just didn’t give you all the info you truly needed.
AI is terrible with ambiguity and conditional data.
Technically it’s completely edible, insofar that it’ll only give you nausea / stomach cramps and a wicked high. Whereas white amanitas are lethal.
The “properly cooked” here refers to well dried and sort of cured material, which has more uh, I want to say “muscimol in relation to ibotenic acid”, iirc. Your liver will also convert the ibotenic acid into muscimol, but that’s where the nausea would come from, as your liver works hard and there’s metabolic byproduct or some such.
But when you properly dry the shrooms, a lot of that ibotenic acid gets tuned into muscimol, which doesn’t usually cause nausea that much.
Skynet takes this as an insult. Next you’ll imply that it’s an Oravcle product.
Amanitas won’t kill you. You’d be terribly sick if you didn’t prepare it properly, though.
Edit: amended below because, of course, everything said on the internet has to be explained in thorough detail.
Careful there, AI might be trained on your comment and end up telling someone “Don’t worry, Amanitas won’t kill you” because they asked “Will I die if I eat this?” instead of “Is this safe to eat?”
(I’m joking. At least, I hope I am.)
Amanitas WILL kill you, 100%, No question.
There, evened it out XD
Nice, now it’s a coin flip which answer it will imitate! ;-)
Yeah, thinking that these things have actual knowledge is wrong. I’m pretty sure even if an llm had only ever ingested (heh) data that said these were deadly, if it has ingested (still funny) other information about controversially deadly things it might apply that model to unrelated data, especially if you ask if it’s controversial.
They have knowledge: the probability of words and phrases appearing in a larger context of other phrases. They probably have a knowledge of language patterns far more extensive than most humans. That’s why they’re so good at coming up with texts for a wide range of prompts. They know how to sound human.
That in itself is a huge achievement.
But they don’t know the semantics, the world-context outside of the text, or why it’s critical that a certain section of the text must refer to an actually extant source.
The pitfall here is that users might not be aware of this distinction. Even if they do, they might not have the necessary knowledge themselves to verify. It’s obvious that this machine is smart enough to understand me and respond appropriately, but we must be aware just which kind of smart we’re talking about.
Do you have a source on them not being able to kill you? Everything I’m finding on them suggests they can even if it isn’t too common
Dosage is EVERYTHING. If you’re slamming a bunch into your gut, you will die. For example, one night I did 8-9 grams of psilocybin (combined with DMT and a touch of acid - suffice it to say, I broke through the edge of the universe and looked back in dismay). Half that in Amanitas would probably kill a person.
HOWEVER, since this is something that seems to come up a lot, here are the potential side effects and a proposal to limit sales of these things due to dumbasses like the guy described below.
https://www.ajpmonline.org/article/S0749-3797(24)00163-6/fulltext
MUSCARIA
Increased consumer interest and widespread availability are concerns for public health because Amanita muscaria products contain compounds that are toxic including muscimol, ibotenic acid, and muscarine.9 Most scientific literature on the health effects of Amanita muscaria in humans pertains to studies of the ingestion of raw Amanita muscaria mushrooms. These effects include dizziness, dysphoria, visual hallucinations, agitation, ataxia, muscle fasciculation, seizures, and coma.12 While death is rare, it has been reported as an outcome, including a case reported in the last year of a 44-year-old man who died after ingesting 4 dried Amanita muscaria mushroom caps.13 The first documented case of hospitalization due to Amanita muscaria consumption in the United Kingdom was reported in July 2023.14 This case involved a 46-year-old woman who had ingested dried mushrooms (0.5 grams) daily for 2 weeks as part of what is referred to as a “microdosing” regimen that was being followed in an attempt to reduce anxiety without inciting psychotropic properties. She reportedly purchased 20 grams of Amanita muscaria mushrooms from a website advertised on social media.
You dry the red ones and this makes something in them less bad for your stomach.
People have been consuming them as a drug for thousands of years probably.
Other anamitas are maybe poisonous, don’t know
Amanitas phalloides, the death cap, is 100% deadly. As are the destroying angels, another group of Amanitas sp. (bisporigera, ocreata, virosa). Amanitas muscaria is 100% edible with proper preparation. The more you know…
There is a youtuber that ate some, he also talks (and tries) a lot of natural drugs. While I’d never recommend someone do it, almost anything is toxic in the right dosages (even water).
