For context, this is related to a article posted on a sub stack called the rationalist which made efforts to disguise the fact that it was GPT generated:
https://news.ycombinator.com/item?id=34287747
The author, Petra, defended themselves saying that English wasn't their native language but that the "seed" content was their own. This article would appear to indicate that they also plagiarized the original source material.
I abandoned my faith once I was unable to rationally reconcile secular and divine morality, which was a theory I played around with in college and later wrote a paper about.
I grew up on DOS 6.22 and quick basic 4.5, so I have a huge amount of nostalgia for how low friction it was to start programming back in the day.
There's a scene in Stranger Things where Sean Astin's character tries to brute force a code using a basic variant, and it was surprising to me to see code on a popular TV show that was remarkably faithful to the original.
Modern IDE's are finally approaching a "low friction" workflow with things like LSP and DAP, which are largely independent of the programming language. The intuitive UX of QuickBasic, VB-DOS, the Turbo series etc. however is still very compelling, in a way that has yet to be comprehensively ported to modern systems.
A few years ago I started taking up running regularly but I didn't have a buddy that I could run with so I decided to adopt a dog. Unfortunately the shelters were relatively far away from my home making it inconvenient to go in and check continuously on the weekends. I later realized that nearly all of them connected to a central city database which contained all of the dogs in the nearby shelters and furthermore that that database was behind a publicly available rest endpoint.
I whipped up a program that periodically every 10 minutes would query the database filtering for dogs matching the qualifications I was looking for using a combination of regex to search for any dogs that were:
- Between 50 and 100 lbs
- Did not have a history of behavioral problems
- Matched a list of active breeds
When it found a potential match it would send an SMS via Twilio to my phone with a picture of the dog, a link to the shelter, and a picture.
Several years later my huskee / Pyrenees hybrid is the best running partner I could ever have asked for. That's my proudest hack and I think she would agree with me.
Have to let the AI experts speculate on why SD goes nuts there because it definitely knows what "The Godfather (1972)" means (if you ask for e.g. 'A still of Patrick Stewart in "The Godfather (1972)"' you get one - which I believe DALL-E can't do because of their facial restrictions?)
It depends on what you're generating, complex prompts in DALLE ("a witch tossing rapunzels hair into a paper shredder at the bottom of the tower") blow midjourney and stable diffusion out of water.
But if all you're doing is the equivalent of visual mad Libs: "Abraham Lincoln wearing a zoot suit on the moon.", then SD and MJ suffice.
With several thousand images on each, I agree with this -- to a degree.
Dall-E does seem more aware of relationships among things, but using parens and careful word order in some of the SD builds can beat it. By contrast, even most failed images from MidJourney could still be in an outsider art gallery. MJ aesthetic works, while Dall-E seems like a 9 year old was taken hostage and clipped out Rapunzel and the paper shredder from magazines and pasted them onto a ransom note.
That said, I have not been able to get any of Dall-E, MJ, or SD to give me a coherent black Ford Excursion towing a silver camping trailer on the surface of the moon beneath an earthrise.
At cost per image, I could pay to get complex concepts such as this rendered via any number of art-for-hire sites at less expense and guaranteed results.
Only some builds support it. This is the one I'm familiar with: [0]. () around a word causes the model to pay more attention to it, [] around a word causes the model to pay less attention. There's an example at the link.
We should get in touch. I built a language partner in Chinese about a week ago using GPT, continuous real-time speech recognition via webkit speech recognition, and Polly TTS so you don't even have to press any buttons, you can just have a fluid conversation the entire time.
Building a Pictionary multiplayer AI game with stable diffusion https://specularrealms.com/2022/10/04/stable-diffusion-picti...
Sean Aston uses BASIC to Hack https://specularrealms.com/2021/04/21/strangest-things/
How free will is constrained from a physiological standpoint
https://specularrealms.com/genetic-freedom/