Hacker News Reader

Comment by bemmu | Hacker News Reader

I was supposed to be making a video game, but got a bit sidetracked when DALL·E came out and made this website on the side: http://dailywrong.com/ (yes I should get SSL).

It's like The Onion, but all the articles are made with GPT-3 and DALL·E. I start with an interesting DALL·E image, then describe it to GPT-3 and ask it for an Onion-like article on the topic. The results are surprisingly good.

View on HN

Thanks, finally a legit news publication :)

This was really funny :)

http://dailywrong.com/man-finally-comfortable-just-holding-a...

Somehow these articles are more readable than typical AI-generated search engine fodder... Is it because I'm entering the site with an expectation of nonsense?

Probably because, by the creator's own admission, the articles are heavily cherry-picked to make sure the output is decent, which is probably a lot more human effort than goes into the aforementioned search engine fodder.

http://dailywrong.com/sample-page/

I would guess that most Spam farms are not using openAI davinci model which is really really good, but expensive. Just a guess.

That's... so weirdly ironic I can't even! Blogspam websites are made by real humans with little oversight, while a literal AI with oversight generates better results.

That said, with a little tweaking, these technologies can - and probably already are - being used to churn out blogspam websites left and right, fully automatic.

So the other men in the pictures are the uncomfortable ones?

Yes, I actually LOLed at that one!

Feels like the headlines could be generated similar to the style of "They Fight Crime!"

"He's a hate-fuelled neurotic farmboy searching for his wife's true killer. She's a tortured insomniac snake charmer from a family of eight older brothers. They fight crime!"

https://theyfightcrime.org/

Here's an implementation in Perl.

http://paulm.com/toys/fight_crime.pl.txt

lol that site is great

>He's an unconventional gay paranormal investigator moving from town to town, helping folk in trouble. She's a violent motormouth wrestler from the wrong side of the tracks. They fight crime!

>He's a Nobel prize-winning sweet-toothed rock star who believes he can never love again. She's a strong-willed communist widow with a knack for trouble. They fight crime!

>He's an obese white trash barbarian with a secret. She's a virginal thirtysomething traffic cop with the power to bend men's minds. They fight crime!

I definitely want to see the DALLE illustrations for this!

The results with things that are artworks or more general concepts are fascinating, but there is for sure something creepy with "photorealistic" human eyes and faces going on...

If you want to see some really creepy AI generated human "photo" faces, take a look at Bots of New York:

https://www.facebook.com/botsofnewyork

Unfortunately the content of that project is a hostage of Facebook now - similar to ransomware gangsters they force you to do something to get the data, in this case you need to create an account and take part in that global surveillance network. I do not understand why people do that.

Hyperbole will get you nowhere good (Just ask RMS).

How about wording your comment in a way that highlights why it’s a shame these pictures aren’t accessible for those without a Facebook account, and skip the whole “you’re murdering puppies” bit?

Being a creepy sex pest will get you nowhere good (Just ask RMS).

The hyperbole is good marketing for a certain audience.

They intentionally prevented it from being able to general realistic faces to reduce the potential for deep fakes

From the server IP looks like you're on some managed WordPress hosting that only offers free SSL on the more 'premium' packages.

Easiest way for free SSL would be to just throw the domain on CloudFlare :)

http://dailywrong.com/new-course-teaches-guinea-pigs-househo...

lol

We joke about it, but an early and very cheap robotic floor cleaner I had was one of those weasel balls constrained in a flat ring harness with a dusting cloth underneath. It was entertaining and not completely useless.

Put a guinea pig in there and you'd get the same effect.

Actually got a chuckle out of the duck one (http://dailywrong.com/man-finally-comfortable-just-holding-a...). Thanks! I hope your keep generating them. Kind of wish there weren't a newsletter nag, but on the other hand it adds to the realism. Could be worthwhile to generate the text of the nag with gpt too; call it a kind of lampshading.

The part where you have to confirm you are not a robot to subscribe to the mailing list is the best part of this, my new favorite website.

Spam advertising is about to reach whole new levels of weird.

Haha, I was in a very similar boat when I built https://novelgens.com -- I was also supposed to be making a video game, but got a bit sidetracked with VQGAN+CLIP and other text/image generation models.

Now I'm using that content in the video game. I wonder if you could use these articles as some fake news in your game, too. :)

This is clever. Does GPT-3 come up with the title of the article, too? That's the funniest part.

At first I came up with them myself, but found that it often comes up with better ones, so I ask it for variations.

I think I got it to even fill the title given a picture, something like “Article picture caption: Man holding an apple. Article title: ...”. Might experiment more with that in the future.

How do you prompt GPT-3 to come up with the titles? That’s an interesting problem.

Well, then I'm impressed with GPT-3's ability to generate those titles!

The combination of photo/title feels like they come from the more absurd articles published by theonion.

If we aren't living in a simulation, it's just a matter of time...

http://dailywrong.com/wp-content/uploads/2022/07/DALL%C2%B7E...

Hot dang. Some Reddit subs can be auto-generated now.

It's already a thing:

https://www.reddit.com/r/SubSimulatorGPT2/

but yea, it will have generated images now.

Hmmm, these seem less deranged than headings from the previous Markov-chain bots—and kinda less interesting because of that.

I guess Markov chains for the headings, Dall-E for images, maybe GPTx for comments. And/or the GPT models should be made wackier somehow—less coherent, perhaps.

I had the same thought when I saw this one http://dailywrong.com/wp-content/uploads/2022/07/DALL%C2%B7E...

This is amazing! Honestly one of the first uses of GPT3/DALL E that has held my attention for longer than a few seconds.

Hilarious. Those first images have hurt feminism more than the GOP has in 30 years.

http://dailywrong.com/wp-content/uploads/2022/07/DALL%C2%B7E...

Parenting > "Gillette Releases a New Razor for Babies"

I loved how it just consistently decided that if babies have facial hair, it's always white fluff.

I think it's because it's using images of babies with soap on their face to learn. Still funny though!

Haha I love this one: https://dailywrong.com/why-your-dog-could-be-an-alien/

Very funny! The "Scientists Warn New Faster Toothbrush May Cause Insanity"-story is not fake though, I've experienced it ;)

This a fucking fantastic site, it’s absolutely hilarious, and I’ve bookmarked it - I kinda unironically want to set it as my home page - but just a heads up that the CSS is broken for me on my iPhone SE2.

The images don’t scale properly with the rest of the site, they’re massive compared to the content.

These results are pretty amazing. Are these cherry picked / curated / edited at all?

Yes they are heavily cherry picked. The web site itself has a disclaimer about it.

This one seems like it could actually be real in Japan: http://dailywrong.com/anime-pillow-gym-opens-in-tokyo/ ;)

It would be interesting to see if there was a market for a monthly newsprint version similar to the old https://weeklyworldnews.com/

Can DALL-E render Bat Boy?

I’m curious, if they’re only making DALL-E accessible now, and if GPT-3 was never really accessible (as far as I know). How do you have access to these things to generate text and images?

There's no waitlist for GPT-3 now. DALL-E is an unrelated product; you don't need access to both.

DALL-E was accessible by invite/waiting queue. GPT-3 is available for pretty long.

You should let readers rate the articles. This way readers new to the site can read the best ones first and get a good impression.

Love it! Better than other news I get to read these days. Some of it rings..like the bluebird suing the cat.

Thank you! Bookmarked!

These are actually quite funny. A bit of a surreal touch, but that makes them even more fun.

Your images are coming over SSL - so won't show up on many browsers (E.g. Firefox)

I've had that idea since GPT 3 but never got any access...

Try Auto-Install Free SSL plugin, it easy for me

This is great, I love it.

Why do the images load so slow though?

This is fantastic, the fake news the world needs.

NGL this shit is pretty cursed and I like it.

This is the most wonderful thing ever.

You get + two million points from me for not having HTTPS.

How do you generate the original image? And what about the subsequent images, do they come automatically from the text? I'd love to know more about the process.

Reply on news.ycombinator.com