The automation of human creativity is advancement to you?

20

Also you used the wrong tense of “advance”.

24

Anonymous OP 20h

go easy on them, they forgot how to write on their own

19

Anonymous #4 17h

They took amazing research and made it into something disgusting 100%

20

Anonymous #5 17h

Generative AI is so awful

15

Anonymous #6 14h

Why should we not be allowed to own certain organizations of 1s and 0s stored as magnetic imprints on spinning disks? This is how you sound, moron.

-2

Anonymous OP 13h

Sure but there’s a difference. People lump all LLM usage into this evil generative AI category. Generative AI is much more broad. If you actually knew what a sparse auto-encoder was, having done machine learning work, you would realize how YOU sound, moron.

5

Anonymous #6 13h

Whereas you have some archaic definition of a hard drive you think disqualifies my point. Not all generative AI is the same. Not all of it has the same purpose. Your “1s and 0s” cover a dubious range of pretty much any software. Generative AI is a category, and it doesnt mean what you think it does, and it doesn’t work how you think it does

2

Anonymous #6 13h

If you aren’t a computer scientist, shut the actual fuck up about what I’m allowed to calculate and what I’m not.

2

Comment Deleted

0

Anonymous #7 13h

Frrr. Like tell me what a logit is. Go.

4

Anonymous #6 13h

The sheer number of people who don’t understand the technology whatsoever make the worst arguments. At least #6 tries to make an argument whereas OP just wants money in an AI company’s IPO. The technology can only benefit if used correctly. Generative AI isn’t going to destroy the world. It may not be the most preferred form, but it’s most certainly not done evil amalgamation many uninformed individuals make it out to be

5

Anonymous #6 13h

I love learning new things about it bc it has the potential to be something really beneficial (AI in general). It’s a bleeding-edge technology. To not want to learn more is almost naive

2

Are you for real right now? Who do you think discovered sparse autoencoders? How do you think a next-token-predictor becomes a part of an agentic dynamic?

1

Anonymous #6 13h

Discovered their usage with neural networks**

1

Anonymous #6 13h

You have plenty of opinions about things you don’t fully understand, as do I, as does everyone. We don’t have to be experts on a topic to make informed decisions about it. And what’s informing me are the facts and the effects that generative AI, especially LLMs and image generators, are having on society at large. Generative AI is being used to spread convincing political misinformation. Generative AI is being used to make human artists obsolete. Generative AI is being used to make child porn.

1

Anonymous #6 13h

The tech has been around so long, idk why everyone is so frightened by the idea that we can express concepts and words on a multidimensional matrix. Surprise surprise

1

Anonymous #7 13h

Its hilariously dumb. Arent they the least bit curious how a computer can parse meaning from a string?

1

Anonymous #6 13h

And I’m taking the bait because you genuinely pissed me off. It’d be easy for me to remind you that you’re not even understanding the concept of a hypothetical to begin with. My original prompt was: “If you could make all genAI vanish, would you” Your point was: “Why shouldn’t I be allowed to own <insert arguably non-genAI system>” Your response is laughably incoherent given the initial question.

6

Anonymous OP 13h

It only sounds that way because ur not letting those words sink in. Ur not taking their meaning and applying it

1

Anonymous #6 13h

Curiosity is not a good reason to create weapons. I think we’d be better off if we never created the nuclear bomb, even if it meant we never harnessed nuclear power. The risks of generative AI far outweigh the benefits, and if a technology’s primary use is detrimental to society, then I don’t see a purpose in continuing to develop and spread it.

0

Anonymous #6 13h

Oh, so you’re NOT trying to hint that because I’m not a computer scientist, I shouldn’t be allowed to have an opinion on generative AI? Because that’s great, I was hoping you weren’t stupid enough to suggest that you need to be an expert on a topic to make informed opinions about it. Because hoo boy, I’d have words for you if you ever dared to have an opinion on something like the economy or healthcare.

0

Anonymous OP 13h

You can have an opinion, it just has to include specifics. What exactly is the harm in a semantic graph?

1

Anonymous #7 13h

“The technology can only benefit if used correctly” is the funniest possible response you could have given. That’s the case for EVERY weapons technology ever invented. And yet, we have things like gun control and nuclear arms treaties. Because inevitably, we DON’T only use technology correctly, and it’s idiotic to try and live in some fantasy land where the military doesn’t use AI to drone strike schools without human input. The risks of genAI are not worth the “benefit” if there even is one.

0

Anonymous OP 13h

Welp we’ve been trying to get to the bottom of where meaning comes for literal millennia. Not every generative ai is a fucking weapon. Yes AI psychosis is horrific and frightening, yes climate change is real, and yes, many are complicit with plagiarism… none of this is inherent to the technology

1

Anonymous #7 13h

The difference is that the maximum potential benefits of generative AI amount marginal improvements to protein predicting models (look at the lack of a difference between non-genAI Alphafold vs the newest version) while the maximum potential benefits of nuclear technology is nearly limitless energy for everyone. Some things are just not worth it, and the vast potential for harm that AI is unjustifiable, in my opinion.

0

Anonymous OP 13h

That is generative ai…

1

Anonymous #6 13h

Again, you could say the same thing for FUCKING EVERYTHING. And yet we have regulations or outright bans on those things. Even fucking eugenics could make literal superhumans paired with modern genetic engineering, but it’s totally banned because of how great of a potential for harm the field has. Under your logic, eugenics is a-ok because it COULD have a massive positive impact on the world, disregarding its even greater potential for destruction.

1

Anonymous #6 13h

Does any of this make any sense to you? Or are you incapable of making comparisons between similar concepts as well as being unable to understand hypotheticals?

1

Anonymous OP 13h

Look dude, ur cherrypicking. Saying Alphafold 3 (a generative ai, as u mentioned) wasnt a HUGE improvement is downright disgusting

0

Anonymous #6 12h

No, it doesn’t. You don’t need to know how solid-phase oligonucleotide synthesis works to know you shouldn’t make an unborn baby faster and stronger by replacing their endogenous ACTN3 gene with an engineered version.

4

Anonymous #6 12h

That’s the same as asking what the harm is in owning the lower receiver to an AR-15, despite that piece on its own being unable to fire a bullet. Components of a larger weapon are still dangerous.

1

Anonymous OP 12h

It is absolutely true that you do not need to be a mechanic to know that a car crash is bad. But if you specifically claim that a spark plug is harmful, you need to be able to explain what a spark plug is.

1

Anonymous #6 12h

It’s also downright correct. Maybe there’s something to your claim that you shouldn’t have opinions on things you know nothing about. Elofsson A. AlphaFold3 at CASP16. Proteins. 2026 Jan;94(1):154-166. doi: 10.1002/prot.70044. Epub 2025 Aug 25. PMID: 40851426; PMCID: PMC12750027. And explain how exactly I’m cherrypicking? Because if you can think of greater benefits to generative AI than its mediocre medical applications I’m all ears.

1

Anonymous OP 12h

Also… you could stretch that logic so far. We should ban popcorn because corn syrup is bad for you.

1

Anonymous #6 12h

I actually agree with you here. But what you’re asking me to do is the reverse. You’re asking me what the harm in a spark plug is while I’m saying that car-centric infrastructure is bad.

1

Anonymous OP 12h

So actually, ur even cherrypicking within the very same research papers. That very same researcher talked about how ONLY Alphafold 3 could model non-proteins. Furthermore, good luck modeling the complex chemical interactions exclusive to 3 that have accelerated pharmaceutical advancement.

1

Anonymous OP 12h

Not to mention video to speech for people who can’t hear, and unprecedented speeds when it comes to detecting vulnerabilities in security systems

0

Anonymous #6 12h

Quite frankly, I didn’t bother mentioning non-protein interactions modeled with Alphafold because they’re a curiosity at best. Molecular interactions still fall squarely in the wheelhouse of non-generative molecular dynamics simulations. Alphafold interactions only provide the roughest of rough estimates when it comes to interactions, and frankly, it doesn’t do anything that motif analysis software didn’t already do.

1

Anonymous #6 12h

AI has uses for accessibility, I admit, but in terms of cybersecurity I’d argue that AI is, at best, a neutral development. It’s just as much, if not more so, a weapon to exploit vulnerabilities as it is a tool to patch them. And again, I’ll draw your attention to AI greenlighting drone strikes on innocent people. Accessibility isn’t worth the deaths it can cause.

0

Anonymous OP 10h

Ummm… DNA??? RNA??? Also, AlphaFold 3 has HEAVILY outperformed any traditional physics-based docking software in structural accuracy

1

Anonymous OP 10h

Also, your last statement about there being no difference in capability compared to motif analysis is unequivocally false

1

Anonymous #6 10h

I don’t know where you’re getting your information, because you’re just plain wrong. Physical simulations far outperform any AI model at, well, predicting physics. And I’ve seen Alphafold’s predictions of DNA-protein interactions, and every single time they’re brought up, it’s at the end of a talk with big triple asterisk saying and the presenter says, “this is just speculation based on preliminary data”. Motif analysis does the same exact job, but it just doesn’t produce pretty looking models.

1

Anonymous #6 10h

Idk why you’re even fighting me on this when you clearly don’t know anything about molecular biology. It’s hypocritical for you to even have an opinion. And there is a difference between Alphafold and motif analysis. Motif analysis does make models. They both provide only preliminary data that needs to be followed up on experimentally. There will never be a substitute for physical data.

1

Anonymous #6 10h

Doesn’t make models***

1

Anonymous #6 10h

I’m not saying Alphafold isn’t useful. It’s a decent intermediary if you want to see if the interactions predicted by motif analysis end up looking like anything interesting, but calling it some sort of revolution is just wrong. Scientists just don’t really consider it anything but a minor component of a much larger experimental pipeline, even in structural biology. Because the #1 thing that everyone asks every time someone shows an AF model is this: “Did you follow up on this experimentally?”

1