This reminds me of a short story by Ken Liu (https://www.uncannymagazine.com/article/50-things-every-ai-w...). He trained an AI on his own corpus of stories and made it spit out a list, then he wrote a story around it. Favorite item on the list: "I never expected to sell my rational numbers."
I made an AI write a story about AI then I made another AI illustrate it
181–190 of 201 posts
Re: I made an AI write a story about AI then I made another AI illustrate it
#182Earlier quoted context omitted.
I believe the point of the original comment is to make AI that comes up with its own original style, not one that "reinvents" an already existing style after first being trained on a different already existing style.
Painters get feedback from the public. Would the AI get the same signals? Then it would be able to evolve new styles based on that.
1) I question this premise. People who revolutionize a field don't do this with feedback from the public. They do it by having a very unique perspective that they bring into being by sheer force of will - usually against great pushback.
2) If we assume your premise is true. The public does not have a preconceived notion of what the art should be like. Presumably you would show them a style of art that they have never thought about before and they will respond positively to it.
Re: I made an AI write a story about AI then I made another AI illustrate it
#183Earlier quoted context omitted.
This is amazing. “We use robot dogs to plant the fruits”
> This is amazing. It's Kafkaesque source material not withstanding, it was able to accurately emulate two of the most robotic sounding popular 'influencers,' with tons of ocntent to piece it all together, but and this is a caveat, the only realistic thing that came from it was when they briefly spoke over each other. Everything else seemed obviously fake. Consider this sounds nothing like the Elon you see on the mos…
"It's like a dog that taught itself to speak to humans, but people are complaining that its English isn't perfect."
- someone, on the state of ML (Can't find the source, because Google hates me.)
Re: I made an AI write a story about AI then I made another AI illustrate it
#184Earlier quoted context omitted.
Painters get feedback from the public. Would the AI get the same signals? Then it would be able to evolve new styles based on that.
1) I question this premise. People who revolutionize a field don't do this with feedback from the public. They do it by having a very unique perspective that they bring into being by sheer force of will - usually against great pushback. 2) If we assume your premise is true. The public does not have a preconceived notion of what the art should be like. Presumably you would show them a style of art that they have never…
I don't agree. For example could Shostakovich's symphonies be composed in a world without WW2 and Stalin? They were major factors influencing what people value in an artistic experience, as well as influencing the composer directly.
Bach composed for the religious people, Mozart for nobility / people who could go to the opera, Beethoven for the romantic middle class who owned a piano, and Shostakovich for the oppressed masses in soviet Russia. The historical context shapes artistic criteria.
You can't make new art styles in a vat (brain in a vat), the process is connected to the world.
Re: I made an AI write a story about AI then I made another AI illustrate it
#185Earlier quoted context omitted.
> It will be AI generated shopping channel dynamically generating products for you to buy on subscription. This one of the solutions to the Drake equation This is rather probable business model for non General-AI targeted at the end consumer, incidentally it too reads like an AI response: specifically the part about the Drake equation and reminds me of Carbon Based Lifeforms - MOS 6581, whose lyrics also sound like a…
We are using ML and the pointy end of capitalism to create traps for the human mind (at scale). This is dangerous.
> We are using ML and the pointy end of capitalism to create traps for the human mind (at scale). This is dangerous.
That remains to be seen, if the extent of ML thus far can be measured by it most ubiquitous use case then it's only pitfall is making people worse spellers than they already were. Autocorrect is a form of ML that quite honestly displays the MANY shortcomings at something that has been around for what? Nearly 20 years now?
While concern is the main response to new things, I'm cautiously optimistic, I'm just finding that it's use-cases seem trivial more than impactful let alone existential.
Re: I made an AI write a story about AI then I made another AI illustrate it
#186Earlier quoted context omitted.
That just sounds horrible to me. Breaking Bad was interesting precisely because it had to end. It was about very specific characters, their changing motivations and development. Rebooting every episode just because the premise is kinda cool and someone wants more of it diminishes what the series achieved both visually and conceptually.
I think once the original episodes are done you need to do something different. They went back in time and did a good series on the criminal lawyer. Now that is done. If you could watch 100 different mini action movies where anything could happen. Everyone could die or everyone becomes nice and gets a desk job. It is less predictable and would be a fun thing. What about you want to watch Law and Order SVU but just an…
I do not think different is a good qualifier for good television. As much as I'd like to see more about the character Mike for example, I would not like to see him in some action movie. His character shines in it's very personal moments, while his prowess regarding violence etc. need only to be hinted at.
The action laden sequences are by far the least interesting aspect of Breaking Bad, serving at best as a somewhat believable and tension-relieving climaxes for emotionally taxing, morally difficult and thrilling parts of the show.
The team behind Breaking Bad managed to tell a cohesive, character driven story despite all the creative restrictions making a show for a large TV network causes (think profitability, playing to as broad of a set of sensibilities as possible).
Telling a thousand different stories in the same world and with the same characters drains them of believability and lessens the emotional impact of the stories significantly - they become arbitrary.
Better Call Saul was a good show because it divorced one of the more important characters (but by far not a main character) almost entirely from the original and focused heavily on humor, with light aspects of drama. Breaking Bad was the other way around.
Therefore, I cannot think of a show that is running for it's 21st season as anything more than at best the visual equivalent of easy listening music, at worst a continued cash grab by the studio producing and network distributing it. For a work of art to be meaningful, it needs to come to some kind of conclusion regarding it's content lest it reiterates the same points again and again, becoming boring in the proccess.
Of course, there is something to be said in favor of easy listening music or TV productions intended mostly for basic entertainment, I recognise this discussion is largely predicated on taste. But I think calling something like that art misses the point that has made art historically so important for humans: The intentional representation of thoughts and experiences of the artist(s) in a specific medium meant to elucidate active engagement with the topics at hand.
EDIT: Added a word for clarity.
Re: I made an AI write a story about AI then I made another AI illustrate it
#187Earlier quoted context omitted.
Painters get feedback from the public. Would the AI get the same signals? Then it would be able to evolve new styles based on that.
1) I question this premise. People who revolutionize a field don't do this with feedback from the public. They do it by having a very unique perspective that they bring into being by sheer force of will - usually against great pushback. 2) If we assume your premise is true. The public does not have a preconceived notion of what the art should be like. Presumably you would show them a style of art that they have never…
The fact that we can say a person revolutionized the field is specifically because their work, at some point in time, got recognized by a larger audience. The fact that an artist had pushback is also context that is used to determine something is revolutionary; you cannot have subversive art if there is nothing to subvert.
Re: I made an AI write a story about AI then I made another AI illustrate it
#188More longform uses of DALL-E 2, many more impressive than OP: https://www.reddit.com/r/dalle2/comments/uw43ow/dalle_2_long...
Re: I made an AI write a story about AI then I made another AI illustrate it
#189I made an AI write a podcast transcript then I made another AI read it :D https://soundcloud.com/jn2022/lexman-artificial-podcast-epis...
Did the AI also come up with the topic? or was that prompted?
It's prompted by a premise of the podcast that I wrote and a list of keywords and guests. I also sometimes prod it during the transcript generation to make it more interesting, the ending for example was me injecting a line "I'll leave you with this bit of Vogon poetry".
Re: I made an AI write a story about AI then I made another AI illustrate it
#190I made an AI write a podcast transcript then I made another AI read it :D https://soundcloud.com/jn2022/lexman-artificial-podcast-epis...
How much of the scripts did you handwrite? Very impressive!
Thanks! It's mostly written by the AI but I prod it along at critical points to make it more interesting