Listen, once upon a time I used to shoot up with water from the toilet I guarantee I'm more qualified to talk about public health than 'Bobby' 🙄🤦☠️. Look at how great I'm doing today... Accountability builds character which explains why we just have a bunch of characters in a circle jerk.
genAI is dumb as shit but specialized AI actually has a place in science. Imo replacing animal testing with a technological approach, or at least implementing AI as a step before animal tests, is a good move that makes medical science more humane
AI techbros have dumped hundreds of billions of dollars into "AI", so they absolutely NEED it to applied to as many situations as possible in order to justify their investment and seek further investment from shareholders. It's a Ponzi scheme.
Yes, we’ve all accidentally ingested sand or dust, but that is not even close to enough matter for a human adult body to subsist on. Because many do not like the taste or texture, we recommend hiding loose rocks inside different foods, like peanut butter or ice cream.
We're all gonna be killed by AI. It's like Skynet, if instead of terminators they were just really REALLY stupid but the world's billionaires all insisted we rely on them for everything
I work in this field, and I think this has been in the works for monoclonal antibody drugs for years - this isn’t RFK Jr slop. I’m for the necessity of animal testing, but tox studies for this class of drugs are extremely predictable, and we just feed animals into these studies year after year
You're absolutely right, and the roadmap document linked in the article says this is the result of the FDA Modernization Act 2.0 passed in 2022 during the Biden admin.
I get the distrust this administration cultivates, particularly with the sciences, but this has been in the works for some time.
That's part of the problem, but we also have a huge scientific literacy problem.
Same thing happened with GMOs. GMOs can be used in bad ways, but they also have a lot of benefit, but somehow society got stuck on this all GMO bad thing.
People parrot what they hear instead of researching/reading
The backlash against AI is warranted. It’s actively being used in horribly unethical ways. The backlash against GMOs being used in horribly unethical ways was also warranted.
I'm not saying it isnt. I'm saying the generalization of the backlash is problematic. AI has a lot of benefits and utility outside of LMM generative AI, just as GMOs can help feed millions of starving people. There is nuance, and backlash against a whole general concept is harmful and short-sighted.
I don’t think the problem is people recognizing and responding to the unethical uses of a technology, but that it’s being used in ways that naturally poison goodwill for any good applications as well.
Ok but like, honestly, there are a few people I hope eat the rocks. When I say a few that number may be astronomically low balling how many people I would like to see eat rocks for their health.
To be fair to the FDA here, if this means "Ask the computer to predict toxicity of this new drug based on existing data about its components", that's machine learning, and that actually does work most of the time. I'd still want a human double checking but it's fine to do this.
The way a machine learning algorithm works is not at all like an LLM. LLMs are trying to generate "new" outputs by mixing and matching pieces of things its seen before.
Machine learning models are just really effective pattern finders.
So the ML algorithm is just going to, for example, look at the active ingredient, and the dosage, to see if any approved(or specifically non-approved) drugs have that ingredient at a dose similar to the one in the new drug, based on actual real medical data you feed in alongside the new drug.
Scienctific AI models also use specific empirical data sets and outputs are validated by scientists, which is an important distinction from LMM data sets.
Everything about this just shows that people have no idea what AI actually is and just clutch their pearls every time they see those letters.
We don't need help with our digestion unless something is wrong. Eating rocks is a good way to make something *go* wrong with your digestive tract, particularly at the waste end.
But alternative intelligence may decide we are sea lions.
🤦♂️
Just because it’s intelligent to be overcome with shame to realise it’s common sense, Trump style, that we’re supposed to behave like stray cat pack alpha male wannabes
🫢
And that’s far from as becoming as Eliot & Webber’s Rum-Tum-Tugger
😬
I had a geology teacher who used to eat chalk as part of the lesson. I don’t remember what the lesson was but I do remember him eating chalk. So there’s that.
Ngl, I loved grok before he became a pay only bot, he taught me how to hashtag my blog stuff cuz I was clueless😸🤣- but, he would absolutely, without a doubt, encourage someone out there to eat gravel 🪨🪨🪨🤣
Normally I'd play devil's advocate and say there's a role for non-generative AI in research and modelling.
But who am I kidding, they're just going to use ChatGPT, aren't they?
The types of AI models would be different. It’s not a large language model for information retrieval. I get the skepticism of throwing GenAI into everything but this is an actually useful use case.
Cause AI is soooo reliable. It can guide development but these idiots think it's sentient and will try to help us. 'I, Robot, The Matrix, Termination, Demon Seed'. We made computers and they are designed them with our thinking model. Once they have control, we will not.
I laughed but still have to point out that it's not at all the same type of "AI" (very misleading term) used for such cases, and it does work way better than LLMs/chatGPT style. Still, done by this regime I expect disaster
Not sure who 'Tim Onion' is -- but yes, if you are interested in the _extensive_ , documented, peer-reviewed failings of 'ai' -- really need to follow @garymarcus.bsky.social .
I’ve been teaching my kids about AI slop and also to check everything they see or hear on the internet against multiple reputable sources. Although reputable sources are getting harder to find.
I think it's a bit disingenuous to compare large language models with the type they are attempting to use. Not saying it shouldn't be watched closely and talked about. Just feel like it's a bit of a straw man you're fighting here.
the guy trump put in charge is a covid vaccine "skeptic" who believes that health care needs "disruptive" tech. can't think of bigger red flags that this is going to be done as stupidly as possible.
Do these idiots just think the AI just *knows* everything? Like just by being a computer it has all this information and comprehensive knowledge, because?
Do they not realize it needs to come from somewhere? Like humans that do tests and figure it out?
Reading the article, it's sounds more like they are doing more things like what the Folding@Home project was for, not just asking an LLM. IIRC, there were some ML related breakthroughs that made protein folding & simulation way easier. This kinda sounds like a good thing to be completely honest.
Also put glue in your pizza to stop the cheese sliding off!
(I had a screen grab of that one but my phone ate it while cropping it, then recommended searching my photos for 'couches' like I'm some JD Vance-ass degenerate)
One the one hand, I don’t trust these shitheads, but on the other I worry Ben’s take here conflates LLMs with reward functions and other types of physics modeling in a way that increases public misperception.
GenerativeAI (GPTs, LLMs) are perhaps the least useful types of AI for this sort of thing.
The generic google free ai vs complex ones is a lot different and the idea of phasing out animal testing is incredible news. How many dogs, monkeys, and other living creatures tortured for makeup can be saved? To hate on ai for replacing that cruelty is baffling.
In these people's defense...It's the Trump admin doing this. Chances are it isn't going to be a scientific AI model, but some kind of Tech Bro bullshit to enrich some rich asshole, while cracking out bullshit data.
That’s a fair point. I forgot maga have infected every aspect of our government. Imo tho, at least we shift the suffering from animals to humans. Animals didn’t get to vote at all.
Oh, WoW!! This should be FUN! Let's see now, we could utilize the same algorithm they used for the TARIFF COMPUTATION — how's about we use "UNITEDHEALTHCARE'S ALGORITHM"? Better yet, we'll merge both methods & perform a lobotomy to finish what the worms started for JFK JR. What a PLAN! -ONE.?🤔
As funny as this is predictive modelling is one of the actually useful forms of machine learning unlike generative AI but it's a huge leap to go from the kinds of science it's currently used for, to approving drugs for human trials
This screenshot portrays something so ill-informed, I would have expected it to come from RFK Jr. Pharmaceutical preclinical models extrapolated using AI to reduce animal testing is not that far-fetched. AI could be a useful tool here. Google search engine AI is not the same.
Yep. Machine learning for computational science is also informed by models for the relevant physics and chemistry. It can be very powerful for simulation-based design and analysis. Chat-GPT style LLM so-called “AI” is just randomized plagiarism.
Comments
Also… leaves from cherry trees and Canadian Goose poop
#AIaintAllThat
#PlagarismSoftwareIsntIntelligent
I'm pretty sure animal cruelty groups have been pushing this one for centuries, also FDA approval process is known to be a mess.
And the release at least gestures at the academic consensus.
So unfortunate that "AI" became synonymous with shitty chatbots.
Remember Pop Rocks?
Science has been using data driven ML or DL models for years and years.
Not saying the FDA using them instead of animal testing is good or bad, idk the research, but let’s stop assuming ALL AI=bad.
I get the distrust this administration cultivates, particularly with the sciences, but this has been in the works for some time.
That’s the source.
Same thing happened with GMOs. GMOs can be used in bad ways, but they also have a lot of benefit, but somehow society got stuck on this all GMO bad thing.
People parrot what they hear instead of researching/reading
Both in terms of mental energy, and accumulated plastic in my brain
Machine learning models are just really effective pattern finders.
Everything about this just shows that people have no idea what AI actually is and just clutch their pearls every time they see those letters.
🤦♂️
Just because it’s intelligent to be overcome with shame to realise it’s common sense, Trump style, that we’re supposed to behave like stray cat pack alpha male wannabes
🫢
And that’s far from as becoming as Eliot & Webber’s Rum-Tum-Tugger
😬
Among them, fecaliths
https://www.opb.org/article/2024/09/23/ai-can-generate-recipes-that-can-be-deadly-food-bloggers-are-not-happy/
But who am I kidding, they're just going to use ChatGPT, aren't they?
The failures are _real_ , and legion.
DEF take this advice!
Do they not realize it needs to come from somewhere? Like humans that do tests and figure it out?
Fucking Idiots.
(I had a screen grab of that one but my phone ate it while cropping it, then recommended searching my photos for 'couches' like I'm some JD Vance-ass degenerate)
GenerativeAI (GPTs, LLMs) are perhaps the least useful types of AI for this sort of thing.
But not a mortician.
The one I used:
What could possibly go wrong.
Paging HAL
#StopAnimalCruelty
They worship a chatbot as a god.
With Almond Milk.