neural networks - AI Weirdness (Page 12)

A weirdly broken AI

By Janelle Shane On April 24, 2020 - 3 min read

The other day someone told me about a program [https://deepai.org/machine-learning-model/text2img] that will generate scenes to match a text description. I’m always excited to test out algorithms like this because the task of “draw anything a human asks for” is so hard that even state-of-the-art results

Trained a neural net on my cat and regret everything

By Janelle Shane On April 17, 2020 - 3 min read

Recently I had my first-ever experiment with training a neural net to generate images, when I trained StyleGAN2 to generate screenshots from The Great British Bakeoff [https://aiweirdness.com/post/613754259190808576/this-is-not-cozy-ai-attempts-the-great-british] . Although recognizable as the baking show, it was a highly distorted, nightmarish version - it would have helped

Nonexistent Easter Eggs

By Janelle Shane On April 09, 2020 - 3 min read

I’ve been experimenting with image-generating neural networks, which look at a bunch of images and via trial and error gradually learn to produce more like them. Or at least “like them” according to the neural net’s own interpretation of realistic, which is usually missing a lot. Last time

An AI's idea of a prank

By Janelle Shane On April 01, 2020 - 2 min read

If you’re a longtime reader of my blog, you’ll know that AIs are consistently terrible at humor. Whether it’s a very simple neural net learning to tell knock-knock jokes [https://aiweirdness.com/post/159132506927/in-which-a-neural-network-learns-to-tell] , or a more-sophisticated algorithm trained on tens of thousands of short jokes

This is not cozy: AI attempts the Great British Bakeoff

By Janelle Shane On March 27, 2020 - 5 min read

I’m a big fan of comforting TV, and one of my go-tos is the Great British Bakeoff. It’s the cheerful clarinet-filled soundtrack, the low-stakes baking-centric tension, and the general good-natured kindness of the bakers to one another. What better way to spread cheer and baked deliciousness than to

How to change a giraffe into a bird

By Janelle Shane On March 20, 2020 - 3 min read

When people study the ways that AI generates and detects images, they have to use something as a test problem. Delightfully, a recent paper [https://arxiv.org/abs/2001.03637] decided to train an AI to transform pictures of giraffes into pictures of birds. Why? Apparently, just to see if

Court cases that sound like the weirdest fights

By Janelle Shane On March 15, 2020 - 2 min read

In the course of running a blog where I train neural nets to imitate human things, I have the fun of sometimes being introduced to things that I didn’t know very much about. Like knitting [https://aiweirdness.com/post/173096796277/skyknit-when-knitters-teamed-up-with-a-neural] . Or crochet [https://aiweirdness.com/post/187489831262/first-there-was-skyknit-now-theres-hat3000]

This neural net knows what smells good

By Janelle Shane On March 06, 2020 - 4 min read

Last week I trained a neural net on 1000 candles [https://aiweirdness.com/post/611219307217797120/what-would-a-candle-inspired-by-your-book-smell] , and soon it was producing scents like Frozen Styrofoam, Volcanoes Comfort, Lemon Lime Decay, and Friendly Wetsuit. We have to just imagine what these would smell like (or in some cases, try not to

Smells like the future

By Janelle Shane On February 28, 2020 - 2 min read

“What would a candle inspired by your book smell like?” This was a question that was going around the other day, and most authors got to answer with something nice. My book [https://t.umblr.com/redirect?z=https%3A%2F%2Fyoulooklikeathing.com&t=ZTZkNThhMDM2MDM5M2Y4MDcwZjI3M2RlN2QxNjQ4MDBlYWQ5NTA5NiwxZmZhMGIxY2QwMzMzMjNkNmY3ZGEzYmE3YjkzMTdmMmRhMmVjOWVk] , however… let’s just

This time I DIDN'T train a neural net to generate candy hearts

By Janelle Shane On February 14, 2020 - 3 min read

I’ve trained neural networks to generate candy hearts [https://aiweirdness.com/post/170685749687/candy-heart-messages-written-by-a-neural-network] before, and the process goes something like this: 1. collect as many existing candy heart messages as I can (which was 366) 2. give them to a clueless neural net that tries to imitate them

Tag: neural networks

A weirdly broken AI

Trained a neural net on my cat and regret everything

Nonexistent Easter Eggs

An AI's idea of a prank

This is not cozy: AI attempts the Great British Bakeoff

How to change a giraffe into a bird

Court cases that sound like the weirdest fights

This neural net knows what smells good

Smells like the future

This time I DIDN'T train a neural net to generate candy hearts

Writing

Subscribe

Recent Posts

Botober 2024

Bonus: "Ignore all previous instructions" gets weirder

An exercise in frustration

Bonus: A unicorn goes downhill

Hidden 3D Pictures

Hidden sheep

Follow

Botober 2024

Bonus: "Ignore all previous instructions" gets weirder

An exercise in frustration