Five Slices #20: AI poetry is more human than human poetry
AI gets a voice and body, Buffalo Sentence, Orphan Crushing Machine, Eldritch Horror
Hello! If you’re new to Five Slices, I share five short stories and ideas from science, history, art, business, culture, and psychology in every post. I also write longer essays once in a while that dive deeper into a single idea. Subscribe here to get it in your inbox:
You can check out the complete list of posts here.
More human than human
“I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes.”
This tweet by Joanna Maciejewska went viral last year with a lot of people being quick to condemn AI intervention in the arts. The thing is, AI already does your laundry – through washing machines. AI also stores thousands of your documents, handles your communication, lets you type using your voice, and do a hundred other things. You don’t think of these as AI because they aren’t marketed that way.
AI isn’t going to take away your jobs by making cheap content that nobody wants. Instead, it’s going to give tools that widen the gap between people who learn fast and those who don’t. Ultimately, it’s the results that matter: people don’t care if a good show comes from a single writer or a writers’ room. But what about AI?
On the surface, it looks like people hate AI-generated content: Yesterday, Sam Altman shared a story written by a “Literary LLM.” Some loved it, some hated it (including me). But in this case, people were already biased that it was AI-written. What about a blind test? When scientists asked people to select their favorite poetry from a collection of human and AI-generated poetry, most non-experts weren’t able to tell the difference. People consistently preferred the AI-generated poetry. This is partly because AI-generated poetry is simple and has more mass appeal.
But there’s more: Another part of the test contained samples created by human poets with the help of AI. It turned out that these poems were more popular than the ones that written without the help of AI. So there’s an edge in using AI as a tool, much the same way that you would use Microsoft Word or programming languages.
Writing researcher David Perell interviewed polymath Tyler Cowen about what type of content and writing will last in an AI-dominated age. Here are some takeaways:
AI is generic when you ask general questions. If you ask more specific, detailed, and unique questions, you get interesting insights. This works better when you combine questions and concepts from multiple fields.
It is better to ask a question and follow up 10 times than type one long paragraph with 10 questions.
What AI can’t replicate (yet) is personal details – biographies, memoirs, personal essays and stories still have a moat because people are interested in other people.
AI trains on data. Humans have an edge over AI in areas where the data is still in their heads. A lot of expertise and “secrets” are stored in the minds of experts. Social networks and mentor-mentee relationships will become more important.
Source: David Perell and Tyler Cowen podcast. It’s quite boring, I don’t recommend it unless you’re really into AI and writing. A much better use of your time is David Perell’s podcast with Ward Farnsworth about writing powerful sentences that last through the ages. I discovered the AI poetry research through
.AI gets voice and a body
In August 2023, my friends and AI found this hilarious video which shows a girl in the future bringing an AI boyfriend home to meet her grandfather:
It’s a new spin on an old concept. In the movie Her, an introvert falls in love with his AI operating system. Blade Runner is set in a world where renegade robots are hunted down because humans see them as pieces of property.
Which makes me wonder: As robots get more and more lifelike, when will people start fighting for robot rights? Google even fired an engineer in 2022 for claiming its AI is conscious when all we had were text-based systems, but now…
Sesame AI has released voice agents that sound eerily like humans, and can learn not just the words and phrases you use, but also your intonations and rhythms.
Clone Robotics introduced the Protoclone android. It has a muscular and skeletal structure similar to humans, and moves like humans do.
General purpose agents like Manus can now do a combination of simple tasks like a research intern. It’s still at a basic stage, but it’s going to get better.
We might eventually get systems combining all of these parts to have AI systems walking among us in human bodies. When? Your guess is as good as mine.
You can watch this video by Fireship to learn more.
Buffalo sentence
Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo.
This is a perfectly valid sentence. What?!
The trick is to understand that buffalo is used in three ways here:
buffalo the animal
buffalo the verb, as in “harass”
Buffalo, a place in New York
The above sentence can be rewritten as:
Buffalo from New York which are harassed by Buffalo from New York themselves harass other Buffalo from New York.
But hey, where’s the fun in writing it that way? (More on this here. I didn’t get it the first time either)
Orphan crushing machine
You’ve probably come across heartwarming news like this:
Then you think about it again. Wait, what? Why is the homeless population of an entire city relying on the random kindness of a kid with bone-marrow cancer? Why isn’t the city taking care of its own? Some more examples:
Insurance executives provide funds for widow who was denied pension. (Why was she denied pension?!)
Lady donates $8,000 to settle lunch debt that was preventing high school seniors from graduating (You don’t let kids graduate until they settle lunch debt?)
Kenny Waters was finally free after being wrongfully imprisoned for 18 years for a murder he didn’t commit, after his sister went to law school and fought his case. (How sweet)
You see the pattern here? It’s admirable that people are doing what they can. But it’s like reading a headline saying “Good man pays $100,000 to save 100 orphans from orphan crushing machine.” Everybody praises the good man, but nobody asks “Why is there an orphan crushing machine?! And why do you need to pay money to save children from being crushed?”
There’s an entire subreddit for this topic if you want to ruin your day.
Eldritch horror
I reread “The call of Cthulhu” by H.P. Lovecraft recently. This is how the story begins:
The most merciful thing in the world, I think, is the inability of the human mind to correlate all its contents. We live on a placid island of ignorance in the midst of black seas of infinity, and it was not meant that we should voyage far. The sciences, each straining in its own direction, have hitherto harmed us little; but some day the piecing together of dissociated knowledge will open up such terrifying vistas of reality, and of our frightful position therein, that we shall either go mad from the revelation or flee from the deadly light into the peace and safety of a new dark age.
The call of Cthulhu defined a genre called Eldritch horror. It’s different from other horror because you don’t get scared by gore, shock, or scary monsters. Instead, the storyteller gives you a glimpse of a vast horror that you can never fully understand which will haunt you because of its immensity.
Think of life from the perspective of an ant which finds a circuit board:
An ant doesn’t start babbling when they see a circuit board. They find it strange, to them it is a landscape of strange angles and humming monoliths. They may be scared, but that is not madness. Madness comes when the ant, for a moment can see as a human does.
It understands those markings as words, symbols with meaning, like a pheromone but infinitely more complex. It can travel unimaginable distances, to lands unlike anything it has seen before. It knows of mirth, embarrassment, love, concepts unimaginable before this moment, and then… It’s an ant again.
Echoes of things it cannot comprehend swirl around its mind. It cannot make use of this knowledge, but it still remembers. How is it supposed to return to its life? The more the ant saw the harder it is for it to forget. It needs to see it again, understand again. It will do anything to show others, to show itself, nothing else in this tiny world matters. This is madness.
Eldritch horror is about learning something terrible and living alone with that knowledge because nobody will ever believe you. It’s about never feeling safe again because you realize that you were never safe before. Even death and madness are no longer escapes.
– Source: Tumblr
I love this genre of storytelling. Lovecraft was the pioneer of the genre but there are other great examples:
Junji Ito’s Uzumaki, about a town where spirals start appearing everywhere and driving its population mad.
Annihilation, about a team of scientists stumbling across an alien lifeform that doesn’t even operate on the same instincts as humans.
Flatland, an “autobiographical” novel written by a square in a world of 2D shapes. The square loses its mind when it meets a 3D sphere. When it tries to explain to other shapes what it experienced, they throw it into an asylum.
The best example is this: If a person from 500 years ago time-traveled to today and saw electricity, cars, trains, aeroplanes, self-landing rockets, the Internet, smartphones, video-calls, X-rays, surgery, microscopes, and Mars landings… just the shock would drive him mad and kill him. Probably.
If you know of any good Eldritch horror stories I might enjoy, let me know.1
We take a lot of magical things for granted because we’re used to them – like the fact that we see the world in high-definition and HD color. My friend
wrote a beautiful article on how neuroscientists figured out the science of color vision. You can check it out here (and subscribe to his Substack!)If you enjoyed reading, hit like and share this post with a friend:
Last week, I shared five of my favorite Substacks that are worth your time. You can check out the complete list of Five Slices posts here.
In Five Slices #9, I talked about knowledge hazards, i.e information which could ruin your life. Eldritch horror is related to one kind of knowledge hazard, which shows you the horror of all the things you can never know for a brief instant.
#1 got me thinking - there are many people in this world whose interior lives span incredible depths, but who struggle to articulate (or articulate with fidelity) to that subjective experience (where if the exact visceral feeling isn't conveyed, it's as good as saying nothing at all). So many thoughts never seeing the light of day, lost to the unfortunate mechanics of human memory.
3 thoughts on that, wrt writing with the help of AI: (a) Remember when James Baldwin said 'You think your pain and your heartbreak are unprecedented in the history of the world, but then you read'? There are words you hear sometimes that save your life. Imagine how many there are that never made it out of the mind's eye because somebody could not put words to those ideas!
(b) I went down a Henrik Karlsson rabbit-hole because of one of your posts (thank you so much for that, btw. His heuristic of iterations and constraints have featured in every single one of my journal scribbles this last week. So good!) and his whole thing about why having a blog as a people funnel-filter is fantastic and
(c) The pains and perils of (an) existence underlying even academic literature (particularly in the social sciences) can turn dry doctrinal arguments so lush. It is easy to commodify or 'academic-ify' social issues when you study it as a 4-credit course, but when you sense the pink flesh of the human experience underneath the writing (for instance with much caste-writing) it can really jolt you. Most would agree that policy and law do nothing until cultures change (think Art. 377 in India) - they are a necessary but not sufficient condition. People write beautifully, the rest of us read (even if with half a mind-block), positions actually change and a whole new culture of empathy about said thing is ushered in! Who woulda thunk.
All this to say the one trite thing that you obviously know, but isn't the potency of writing just incredible? I am hopeful that (despite all the lamenting about AI and boilerplate writing) we will witness more heart-on-sleeve writing simply because it is that much easier to convey yourself with a little Claude heart-to-heart and crafty prompts. You risk little because you're less likely to be misunderstood, and you're rewarded with kindred spirits.
I like this take re. LLM fiction (and re. dog fiction).