hbiden@onlyfans.com wrote: Thu Feb 06, 2025 1:26 pm i think it says more about the usefulness of poetry. so to answer your question, writers care.

Any writer worthy of being published knows it (AI) will never write better than a human, or like a human.

It says nothing at all about the usefulness of poetry. I'm curious to know how you reached that conclusion from what you posted. I know some internet crytpo bros think poetry isn't useful, though.

i wouldn't say the problem started with bitcoin... it's the dumbing down of society. i am happy for anyone smart enough to be "national student poet of the west," but i hope she has a backup plan.

ChatGPT isn’t better at writing poetry. We’re worse at reading it.

February 4, 2025
By Diane Sun
Diane Sun is a sophomore at Harvard University studying philosophy and linguistics.

In 2022, I was named the national student poet of the West, one of the nation’s highest honors for youth poets. During my year of service, I performed my work across the country, including at the White House. Yet I couldn’t tell the difference between T.S. Eliot and ChatGPT.

Apparently, I’m not alone in this lapse of discernment. According to a study published in Nature in November, Americans are more likely to appreciate AI-generated poems than poetry from humanity’s most celebrated authors: Emily Dickinson, Sylvia Plath and, of course, Eliot.

In fact, not only did the participants prefer ChatGPT’s poetry, but they also found it “more human than human.” AI-written poetry was 17 percentage points more likely to be judged as having been written by a human than the actual human-authored poetry. It was also rated more favorably in terms of “rhythm and beauty.”

Poetry has now joined a long list of creative pursuits — essay writing, painting, chess and more — at which artificial intelligence is outperforming authentic intelligence. The intrusion of AI into our most sacred modes of expression has largely been met with disgust, anger and fear. But I don’t blame AI; I blame us.

We’re not reading well — or at all anymore. According to the National Endowment for the Arts, the percentage of young adults who read literature declined by 28 points from 1982 to 2002. Twenty years later, literacy skills remain worrisome, as two-thirds of American fourth-graders do not meet national reading standards.

And it’s not just kids: American adults are struggling to read, too. A 2022 Gallup poll estimated that American adults read two to three fewer books per year than they did before 2016. Meanwhile, ChatGPT was trained on the data equivalent of hundreds of thousands (if not millions) of books, giving it the clear edge when it comes to deciphering and replicating the world’s greatest poets, their critics and their critics’ critics.

AI has placed poetry at risk of extinction. Only 12% of American adults read or listened to a poem in 2022. When we do encounter poetry, it’s often trendy “Insta-poetry” — short and straightforward verses popularized on social media, a la Rupi Kaur. This more digestible form of poetry has its merits, but it doesn’t challenge readers to interpret the subtleties of form, meter and allusion the way traditional poetry does.

People prefer what they can comprehend, and collectively, we’ve lost our ability to comprehend poetry. The researchers behind the Nature study found that AI-generated poems used more obvious and direct language, making them more accessible to nonexperts. Participants didn’t have to struggle to analyze the nuanced shades of emotion and complex metaphors that adorn human poetry.

But poetry is not meant to be straightforward. Reading poetry is supposed to take effort. The problem is that we’ve become lazy readers and writers, and AI is profiting from that. The headaches of formality, connotation, meter, grammar, allusion and intention have caused us to discard our literary minds and trade them for convenience. But difficulty is the crux of poetry. ChatGPT will never be able to slyly admonish a critic mid-stanza, deliberate over the perfect phrasing with an editor or become struck with inspiration in the middle of personal tragedy.

Poets travel the nooks and crannies of their brains, negotiating their lived experiences, their sparks of inspiration and their heartfelt emotions. ChatGPT, on the other hand, is programmed to take the path of least resistance. It is a “stochastic parrot,” as some machine-learning researchers have put it. The words of ChatGPT don’t correspond to reality but to an algorithm linking words and phrases based on probability.

Poetry is meaningful because it is art created through adversity, from the unpredictable and unanticipated. We cannot expect to appreciate poetry without accepting that sometimes it leads us to places we don’t understand. Embracing the difficulty of the abstract — whether that’s through including T.S. Eliot in our curriculums or writing our own emails — is how we retain our humanity. We must confront the complexities that intimidate us, the texts that force us to pause, the obscure allusions that require additional research to understand. The niche, the avant-garde, the eclectic: These are the labors that come with being human.

It has taken GTP-4 1,800,000,000,000 parameters to even come close to replicating the literary prowess of poets born before the invention of electricity. If we directed a fraction of the funding, effort and attention currently focused on AI toward poetry, ChatGPT would not seem nearly as remarkable in the face of human ingenuity.

Even after my year-long poetry fellowship, it would take me a thousand lifetimes before I could come close to amassing the knowledge of ChatGPT. But this also means that when I mistook a T.S. Eliot poem for AI, I found out there was another beautiful poem I had yet to ponder, and a lot more room for me to grow.

^The Marvelization of Society

I really find these “oh look AI wrote a symphony” etc arguments absolute bullshit. Yeah, a computer program was trained to replicate an act of human creativity with data from acts of human creativity.

This is about as insightful as a chimpanzee smoking a cigarette.

Gramsci wrote: Sat Feb 08, 2025 1:51 am I really find these “oh look AI wrote a symphony” etc arguments absolute bullshit. Yeah, a computer program was trained to replicate an act of human creativity with data from acts of human creativity.

This is about as insightful as a chimpanzee smoking a cigarette.

what do you think you're arguing against?
AI is not eliminating artists' jobs,
or it is and we shouldn't worry much about it?

hbiden@onlyfans.com wrote: Sat Feb 08, 2025 1:58 am

Gramsci wrote: Sat Feb 08, 2025 1:51 am I really find these “oh look AI wrote a symphony” etc arguments absolute bullshit. Yeah, a computer program was trained to replicate an act of human creativity with data from acts of human creativity.

This is about as insightful as a chimpanzee smoking a cigarette.

what do you think you're arguing against?
AI is not eliminating artists' jobs,
or it is and we shouldn't worry much about it?

I’m not arguing against, I just don’t think this is what it’s for. AI is great for doing the heavy lifting of “office” work, medical research and diagnostics etc, but replacing human creativity isn’t feasible or desirable because it relies on training data that comes from human beings. Will it innovate? Possibly, but if the limitations are existing works then the innovation is likely to come from churning out thousands of options and a human stopping the wheel spinning and thinking “huh, I haven’t seen that combo before”. So possibly it speeds up the process of genre blending that is how human art evolves… mostly I just think it’s business thinking “great, I don’t have to pay someone to create background music for a video game.”

This isn’t what should interest us in AI and is about a silly as anthropomorphic robots. Robots are tools, making one that looks like a person isn’t the point. Drilling a hole in a side panel is…

PS I work in an industry that’s being impacted by AI, architecture. So far it’s okay for computational design, but that’s heavy lifting not designing. At the moment it’s great for studying a site for something like high volume residential or commercial to understand the capacity, economics etc. But the design part is still very much a collaborative human endeavour. Could you train the AI with every variable that goes into a building? Sure, but there are still things like community consultation and non commercial client engagement which AI can’t do. Try getting AI to design a building in Oxford University, a maze of ancient buildings and incredibly strict planning rules…

hbiden@onlyfans.com wrote: Sat Feb 08, 2025 1:58 am what do you think you're arguing against?
AI is not eliminating artists' jobs,
or it is and we shouldn't worry much about it?

It is most definitely eliminating artists jobs. I dunno at what level it currently is exactly, but that is very much the hope of Hollywood and commercial studios.

Gramsci wrote: Sat Feb 08, 2025 3:36 am

hbiden@onlyfans.com wrote: Sat Feb 08, 2025 1:58 am

Gramsci wrote: Sat Feb 08, 2025 1:51 am I really find these “oh look AI wrote a symphony” etc arguments absolute bullshit. Yeah, a computer program was trained to replicate an act of human creativity with data from acts of human creativity.

This is about as insightful as a chimpanzee smoking a cigarette.

what do you think you're arguing against?
AI is not eliminating artists' jobs,
or it is and we shouldn't worry much about it?

I’m not arguing against, I just don’t think this is what it’s for. AI is great for doing the heavy lifting of “office” work, medical research and diagnostics etc

right. i pleaded with my kids to read this tweet but they're too technoskeptical at this point (20 and 16). "dad, AI is cheating! dad, don't share your medical info! i think my professor knows more than AI!" they are blinded by a misconception/ideology that will take some effort to undo.
https://twitter.com/eigenrobot/status/1 ... 3588749591

but replacing human creativity isn’t feasible or desirable because it relies on training data that comes from human beings. Will it innovate? Possibly, but if the limitations are existing works then the innovation is likely to come from churning out thousands of options and a human stopping the wheel spinning and thinking “huh, I haven’t seen that combo before”. So possibly it speeds up the process of genre blending that is how human art evolves… mostly I just think it’s business thinking “great, I don’t have to pay someone to create background music for a video game.”

This isn’t what should interest us in AI and is about a silly as anthropomorphic robots. Robots are tools, making one that looks like a person isn’t the point. Drilling a hole in a side panel is…

and this is where i'm proud to have technoskeptical kids, so it's a double edged sword. they almost get it.

I’ve got a good friend that probably the smartest person I know. A Medical Doctor, with a physics post grad and PhD in AI, that builds 3D full colour x ray machines as a business. He’s also a radiologist in New Zealand’s public healthcare system and on the hard left.

He thinks he is significantly worse at diagnosing from scans than the AI systems he uses to the point he thinks humans shouldn’t be doing that task.

That conversation kind of blew my mind. His comment was human radiologists are basically doing “bird watching”.