OMGPONIEZ!

Last year I trained a neural net to generate new names for kittens, by giving it a list of over 8,000 existing cat names to imitate. Starting from scratch, with zero knowledge of English or any context for the words and letter combinations it was trying out, it tried to predict what letters might be found in cat names, and in which order. Its names ranged from the strange to the completely nonsensical to the highly unfortunate (Retchion, Hurler, and Trickles were some of its suggestions). Without knowledge of English beyond its list of cat names, it didn’t know what letter combinations to avoid.

So I decided to revisit the cat-naming problem, this time using a neural net that had a lot more context. GPT-2, trained by OpenAI on a huge chunk of the internet, knows which words and letter combinations tend to be used together on the English-language internet. It also has (mostly) figured out which words and letter combinations to avoid, at least in some contexts (though it does tend to suddenly switch contexts, and then, yikes).

When I trained GPT-2 on the list of cat names using Max Woolf’s colab notebook, it still retained a lot of what it had learned from the rest of the internet. Gone were the strange names like “Tilly-Mapper” and “Balllucidoux” - it had a bunch of real words it could use instead. Here are some of the names it came up with - and the Morris Animal Refuge (who you may remember from that time they used neural net names for their guinea pigs) has given some of these names to some highly adoptable kittens.

First, neural net can do fancy:

Taffeta Pompompur Monocle Tom Glitter Notion Tinnitus Cheesemonger M. Tinklesby Linklater Soap

It can also do the opposite of fancy:

Scat Cat Butthole Gangrene Moisture Grotesque Petard Oilbag Buttwig The Cream Meatbag Dr Fart Fudge Putty Scumbag Constipation BUTT

And it can also do names ranging from tough to downright sinister:

Miss Vulgar Lillith The Vamp Elle Fury Deadbolt Romeo of Darkness Starmaker Fist Warning Signs Bibles Smoked The Firestarter Higher Rune Scarlet Be Thy Coat Kill All Humans Bones Of The Master Mr. Sinister Evil Whispers Spawn Serendipitous Kill Stranglehold

(Starmaker and Sparky Buttons are from a litter that had upper respiratory infections that damaged their eyes, but even though their world is kinda cloudy, they love to play and cuddle.)

I’m a particular fan of the Very Weird cat names:

Honeystring Dr Leg Tom Noodle Pinball Scene Peanutbutterjiggles You’re Telling A Lie Beep Boop Thoughts Bobble Bun Atmosphere You Name It Whiskeridoo Sparky Buttons

Seemingly This Guy Various Authors Chicken Whiskey Fish Especially Thelonious Monsieur Ringo Shuffles Sweet Cakes EXTAs (Eye Stalks) Checker Spin Donut Quin Two Patz Grandpa He Glad Funky Moe Fluttering Feelers Accepted A Tribute Chewie Bean PLEASE Gregory Chimney Notable PRODUCT LEGEND Weird Science Platinum Not Suitable For Character the Enforcer

Did I mention these cats are adoptable? If you live near Philadelphia, you live near these kitties!

So the other day I heard from Change.org, a company that lets anyone make an online petition and gather signatures. In over a decade of existence, they’ve hosted about 5 million unique petitions.

Some of the petitions are VERY unique - like the ongoing petition to sell Montana to Canada, which gathered so many signatures that the Montana House of Representatives introduced a bill to release a statement opposing the sale. The bill failed to pass.

The question that Change.org - and I - became obsessed with is: if I trained a neural network on the full list of petitions, what kinds of demands would it generate?

The neural net I ended up using is 117M-GPT-2, by OpenAI, which is better at stringing together readable sentences than some of the other neural nets I’ve used. It also comes with a lot of prelearned knowledge about how words are used in sentences, and how they relate to one another, so it will even suggest things that aren’t in the training data sometimes. As training data to imitate, I gave it about 190k petition titles (heavily filtered for quality). I trained it for several minutes on Google’s free colaboratory GPUs (thanks to a colaboratory notebook put together by Twitter user @roadrunning01), and then played with the sampling parameters (mostly truncation) until it was producing very weird (but still readable) petition titles.

Just like real Change.org petitions, the AI-generated demands were clustered in a few general categories:

Bad ideas/Lost Causes: Dogs are not a thing!! Dog Owners are NOT Human beings!! Help Bring Climate Change to the Philippines! Taco, Chipotle, and Starbucks: Bring Back Lettuce Fries Filipinos: We want your help stopping the killing of dolphins in Brazil in 1970’s Mr.person: I want a fresh puppy in my home Simple Stats Administration: Make Another proboscis. Officials at Prince Alfred Hospital: Aurora to Tell The Company To Send A Baby to Mars Sign Petition for Houston’s New Fireworks to be Offensive

What? Make a mudchat Please not punish myself with a $20 fines. Unicorn: Stop breaking crab products Rooster Teeth : Have Rooster Teeth Fix Your Responses To Obama The people of Great Adventure: get lil bl00ty moose loyal to us The People of Kashmir : Ban of Airbrushed Bamboo Trees By Pune Barack Obama, Barack Obama, and Barack Obama: STOP PING MY HUSBERS! Saskatoon Police Service: No more scootty One Highway, Four Hens, Highway 1 Rhino Amish Culture Association: Cut the horns of the congon sturgeon & treat it better!

Seems reasonable: Harmonix: Increase the speed limit on Easton Road to 5mph. Everyone: Put the Bats on YouTube! Donald Trump: Change the name of the National Anthem to be called the “Fiery Gator” Taco Bell: Offer hot wings and non-perfumed water for all customers Do not attack the unions! Keep cowpies! Anyone: Get a cat to sing on air! The people of the world: Change the name of the planet to the Planet of the Giants Dr James Alexander: Make the Power of the Mongoose a Part of the School’s Curriculum

There must be a story behind this one: Theresa May MP: Stop The Pigeon Rally in Great Britain Karen’s mother: Please bring your own breadsticks for Karen. baseleine planetarium: Unnecessary insults are not welcome in our flat, end it! Anyone: Stop the use of the word ‘shoe’ in a derogatory way. City of Toronto, The City of Toronto, The City of Toronto: Remove the “Bam Bam” sign Basketball Club of St. Louis: Stop the Feral Horseshoes at the Basketball Club Tom Heneghan as OBE Change Council Localimian: Dammit you! Denny: Put one more black bee sweater on Em1nt du Poste Belfast City Council: Fire the cabbages at Clutch MarketI  Kim Hsu: Tougher Penalties for Pedestrians and Elephants on City Street in Austin Texas

This would appear to lay to rest the question of what AI would demand if it suddenly attained sentience. If you are an AI yourself and want to sign on to these petitions, you can register your preference for hot wings and non-perfumed water, and/or put Karen’s mom on notice about the breadsticks.

It is quite fitting that the idea to train a neural net to generate comics names came from the creator of a webcomic about a hapless AI robot. Aaron Uglum collected 1,417 names of newspaper comics from Wikipedia, and I gave them to textgenrnn to see if it could figure out how to make more of them. Then, Aaron gave them these delightful illustrations.

Please enjoy these very strange comics.

I am actively mad that these superheroes don’t exist.

Idea and illustrations by Aaron Uglum - go read his AI robot webcomic, LABS!