Plants, maybe. Fungi, hell no.
AI + fungi = you die
Actually I use it as a starting point for fungi. Seek will usually get me to the genus, and from there I can cross reference various books to narrow it down. Hell, sometimes it'll give me an exact match, and then I just have to perform a yes or no ID with my field guides. That being said, I mostly end up with no, I'm shit scared of all amanitas and most mushrooms just aren't tasty enough to warrant the effort.
I have heard that spore prints are a reliable way of determining mushroom species (removing the stem, putting the underside of the mushroom on an ink pad, pressing against paper, and comparing the print with those of known species).
I bet an AI could analyze that data pretty well. But since there's really no market for such a product, if I want it, I would have to make it myself. In which case I highly advise against using it because I really don't trust me.
There is no ink involved, the spores make the print.
https://doubleblindmag.com/mushrooms/how-to-grow-mushrooms/how-to-make-a-spore-print/
I don't actually know if it's considered a deepfake when it's just a voice; but I've been using the hell out of Speechify, which basically deepfakes voices and pairs them with a text input.
...so... nursing school, we have an absolute fuck-ton of reading assignments. Staring at a page of text makes my brain melt, but thankfully nowadays everything's digital, so I can copy entire chapters at a time, and paste them into Speechify. Now suddenly I have Snoop-dogg giving me a lecture on how to manage a patient as they're coming out of general anesthesia. Gets me through the reading fucking fast, and it retains so, SO much better than just trying to cram a bunch of flavorless text.
Speechify also pays the people who's voices they're using rather than taking them from publicly available videos and recordings without permission.
Wait that's genius. I would listen to Snoop Dogg teaching me particle physics any day of the week.
I think the key here is you're using it for yourself only.
I think it comes down more to understanding what the tech is potentially good at, and executing it in an ethical way. My personal use is one thing; but Speechify made an entire business out of it, and people aren't calling for them to be burned to the ground.
As opposed to Google's take of "OMG AI! RUB IT INTO EVERYONE'S NOSE, THEY'RE GONNA LOVE IT!" and just slapping it onto the internet, and then pretending to be surprised when people ask for a pizza recipe and it tells them to add Elmer's Glue to it...
Two controlled inputs giving a predictable output; vs just letting it browse 4chan and see what happens. The tech industry definitely seems to lean toward the later, which is fucking tragic, but there are gems scattered throughout the otherwise pure pile of shit that LLMs are at the moment.
Do not use ai for plant identification if it actually matters what the plant is.
Just so ppl see this:
DO NOT EVER USE AI FOR PLANT IDENTIFICATION IN CASES WHERE THERE ARE CONSEQUENCES TO FAILURE.
For walking along and seeing what something is, that’s fine. No big deal if it tells you something’s a turkey oak when it’s actually a pin oak.
If you’re gonna eat it or think it might be toxic or poisonous to you, if you want to find out what your pet or livestock ate, if you in any way could suffer consequences from misidentification: do not rely on ai.
You could say the same about a plant identification book.
It's not so much that AI for plant identification is bad, it's that the higher the stakes, the more confident you need to be. Personally, I'm not going foraging for mushrooms with either an AI-based plant app or a book. Destroying Angel mushrooms look pretty similar to common edible mushrooms, and the key differences can disappear depending on the circumstances. If you accidentally eat a destroying angel mushroom, the symptoms might not appear for 5 to 24 hours, and by then it's too late. Your liver and kidney are already destroyed.
But, I think you could design an app to be at least as good as a book. I don't know if normal apps do this, but if I made a plant identification app, I'd have the app identify the plant, and then provide a checklist for the user to use to confirm it for themselves. If you did that, it would be just like having a friend just suggest checking out a certain page in a plant identification book.
The problem with AI is that it's garbage in, garbage out. There's some AI generated books on Amazon now for mushroom identification and they contain some pretty serious errors. If you find a book written by an actual mycologist that has been well curated and referenced, that's going to be an actually reliable resource.
If you're using the book correctly, you couldn't say the same thing. Using a flora book to identify a plant requires learning about morphology and by having that alone you're already significantly closer to accurately identifying most things. If a dichotomous key tells you that the terminating leaflet is sessile vs. not sessile, and you're actually looking at that on the physical plant, your quality of observation is so much better than just photographing a plant and throwing it up on inaturalist
sorry couldn't hear you over the CRUNCHING OF MY MEAL
I have a feeling I know where your username came from.
The blanket term "AI" has set us back quite a lot I think.
The plant thing and the deepfakes/search engines/chatbots are two entirely different types of machine learning algorithm. One focussed on distinguishing between things, the other focussed on generating stuff.
But "AI" is the marketable term, and the only one most people know. And so here we are.
I hate when streamers/gamers/etc refer to procedural generation as "ai generated". It's infuriating.
I particularly "Love" that a bunch of like, procedural generation and search things that have existed for years are now calling themselves "AI" (without having changed in any way) because marketing.
Reminds me of how everything on a computer used to be a "program", but now they're all just "apps"
You're talking about types of machine learning algorithms. Is that a more precise term that should be used here instead of AI? And would the meme work better if it wss used. I'm asking, because I really don't understand these things.
"AI renaming image files for easier search"
You're cool, though, you can stay
Alphafold? Beautiful. Perfect. No notes.
Nudeify type apps? Fuuuuck alllll the way off!
We’re in that awkward part of AI where all the degenerates are using it in unethical ways, and it will take time for legislation and human culture to catch up. The early internet was a wild place too.
I am a physicist. I am good at math, okay at programming, and not the best at using programming to accomplish the math. Using AI to help turn the math in my brain into functional code is a godsend in terms of speed, as it will usually save me a ton of time even if the code it returns isn't 100% correct on the first attempt. I can usually take it the rest of the way after the basis is created. It is also great when used to check spelling/punctuation/grammar (so using it like the glorified spellcheck it is) and formatting markup languages like LaTeX.
I just wish everyone would use it to make their lives easier, not make other people's lives harder, which seems to be the way it is heading.
With all the hot takes and dichotomies out there, it would be nice if we could have a nuanced discussion about what we actually want from AI right now.
Not all applications are good and not all are bad. The ideas that you have for AI are so interesting, I wish we could just collect those. Would be way more helpful than all the AI shills or haters rn.
The difference is that plant identification is a classification problem, not an LLM.
Not all of AI is LLMs, most aren't.
I think state machines are cool and groovy. I still don't understand genetic algorithms but I wish I did.
15 years ago we were all saying "AI is just a series of IF statements" because of expert systems and y'all forgot
The most annoying thing since the rise of LLMs is that everyone thinks that all of AI is just LLMs
Classification machine learning models can also be neural networks, which is something that was called AI also
AI isn't just about LLM. Modern AI libraries (pytorch, tensorflow etc.) can be used for being trained with all sorts of data.
upscaling of old media is pretty cool too
Depends.
Old, niche videogames where the fanbase doesn't have the capacity to do it? Sure. James Cameron replacing Arnold with a UHD leather Muppet in True Lies? Not so much.
I've had to literally perform a Google search to find a customer support phone number before. Because the website of the company just kept redirecting me in circles.
Their phone support was just as useless, though.
It was GameStop, by the way.
Gethuman.com is my go-to. They used to be much better than they are now, but it's still routinely better than trying to navigate automated systems or find phone numbers myself
We need to strike back with an AI customer which alerts us if we could finally talk or chat again with a human if all automatic solutions are discussed.
Using it for plant identification is fine as long as it's an AI designed/trained for plant ID (even then don't use it to decide if you can eat it). Just don't use an LLM for plant ID, or for anything else relating to actual reality. LLMs are only for generating plausible-sounding strings of text, not for facts or accurate info.
First time I've seen this meme template. Love it.
Sign up with iNaturalist for plant and animal identification! Citizen science is good for you.
hmm guess which one also doesn't suck the energy equivalent of a sizeable town
AI search is great.
The more "searchey", and less "generativey", the better. What goes against the direction every provider is going, but it's still great.
I've learned that training a model to search your (companies) unmaintainable, unorganized, and continuously growing documentation storage is a godsend.
1. Be civil
No trolling, bigotry or other insulting / annoying behaviour
2. No politics
This is non-politics community. For political memes please go to !politicalmemes@lemmy.world
3. No recent reposts
Check for reposts when posting a meme, you can only repost after 1 month
4. No bots
No bots without the express approval of the mods or the admins
5. No Spam/Ads
No advertisements or spam. This is an instance rule and the only way to live.
