Twitter's "Art Assignment Bot" Is Generating Prompts For Your Next Masterpiece

Tired of typical, linear art school assignments? Allow us to introduce @ArtAssignBot. Now: "Create a machine denying childhood, due in 53 minutes."

Art school assignments too easy? Your next-level teacher might be on Twitter.

Meet Art Assignment Bot, a Python-scripted Twitter bot whose sole function is its namesake; once an hour, on the hour, it creates a randomly generated art assignment using words pulled from assignments art teachers have used for centuries. It's a process-in-motion that will continue until the Twitter servers are down for good, or until after the bot generates 90,345,024 assignment ideas (which would take approximately 10,000 years).

Jeff Thompson, the artist, professor, and programmer behind Art Assignment Bot, is also responsible for Every Possible Photograph. This custom-encoded software is designed to generate every possible combination of pixels within a low-resolution grid. With an estimated “finishing date” 46 trevigintillion (that's 46 x 1072) years from now, it might be Thompson's longest running process, but Art Assignment Bot is a close second. 

Just this year, another project by Thompson, Computers on Law & Order, was commissioned by Rhizome. For it, Thompson combed through all 456 episodes of the iconic TV show, and took a screenshot of every computer that appears within the episodes. The result is 11,000 images that seem to chronicle an evolution of computer usage parallel to what actually occurred in life off-screen. 

We interviewed Thompson to get some more insight into what goes on in the mind of this modern code artist.

What inspired Art Assignment Bot? Where's there anything in specific that prompted you to create the Twitter account?

Jeff Thompson: Over the past year or so I’ve been making lots of bots, and doing algorithmic writing in general. Bots that ask “would you rather” questions (@wouldratherbot), that traverse an endless dungeon (@dungeon_bot), and a now-defunct bot that replies to random tweets with “Really?” Art Assignment Bot came out of a lot of these kinds of experiments.

I see these bots as creatures or machines, ones that are suited to their ecosystem and produce surprising results or spur interaction between algorithmic text and a human reader. Art Assignment Bot in particular isn’t really all that complicated, but it’s able to produce surprising results and gets people to interact with it. I think that’s the mark of a good bot.

Why did you select the specific thematic art-words (ex: intervention, exploitation, psycho-sexual, etc.) that the Twitter account uses? Were these taken from assignments you worked on in the past?

The “topic” for the assignments are pulled from canonical art school projects and what I see as over-generic or overused art themes (home, family, memory). There are also topics from historical artworks. For example, “war” is based on Goya’s famous “Disasters of War” series. I also threw in some more contemporary curveballs, like "glitch" and what seems to be Twitter’s favorite: “cats on the internet.”

In your mind, should art practice and study be a didactic process? Do you think assignments like the ones from @artassignmentbot actually help improve an artist's skill set? Why or why not? 

I see bots as a form of writing, and as pieces in of themselves. If the tweets engage someone, or make them laugh or think about their practice as an artist or how they look at art, that’s fantastic. I feel more connected to the projects of conceptual art and what Kenneth Goldsmith calls “uncreative writing”. That kind of work could perhaps be called didactic, but I like to think of it as requiring a more active engagement with *how* something is made than what it depicts.

Can you further explain how Art Assignment Bot engages with and critiques traditional art school assignments? Do you have any personal experiences with art assignments like these?

I didn’t really intend the bot to be a critique of art school (I really enjoyed my time there), though it does intentionally try to shake up some topics that I think are a bit over-used. For example, a photograph of a sunset is pretty boring, but these assignments are weird, surprising, and would be challenging to make:

Make a clay form of the history of sunsets, due on Sun, Jun 0.

— Art Assignment Bot (@artassignbot) June 8, 2014

Build a videogame about sunsets, due in 27 seconds.

— Art Assignment Bot (@artassignbot) February 3, 2014

Make a clay form of the history of sunsets, due on Sun, Jun 0.

— Art Assignment Bot (@artassignbot) June 8, 2014

As a teacher, I think the bot does provide some interesting ideas. I would totally love to teach an entire semester dictated by bots and algorithmic systems.

Has anyone actually completed the AAB-generated projects? Can you tell us about any responses to the Twitter account? 

Yes! There seem to be two categories to finished assignments: the quick-and-dirty Internet link and a few actually handmade responses. Here are two recent examples:

@jordan_yearsley @artassignbot that was easy pic.twitter.com/uOec2uJYv1

@artassignbot @evhan55 done pic.twitter.com/XcOTvkJhFc

What other projects are you currently working on?

On the bot side, I have a little one that runs on my personal account that tweets periodically how many files my computer has open at any given moment. I’m interested in a diaristic relationship with technology, especially when it reveals sides we don’t consider. I have another little bot that will go live soon called Better Than Bot that makes random “this is better than that” statements. I’m hoping it sparks virulent arguments!

Aside from that, I’m spending my summer on a few other projects: building a machine that counts grains of sand (to make a sculpture of all the 0s and 1s on my hard drive) and creating a computer-vision algorithm to let computers recognize other computers.

For the code-based out there, Jeff Thompson has graciously made available the script of @ArtAssignBot.

More of Thompson’s projects can be viewed on his website.


2,800 Screens Create Internet-Enabled LCD Mosaic

Kenneth Goldsmith Printed Out 33 GB Of The Internet In Support of Aaron Swartz

God In The Machine: An Interview With Computers Club Founder Krist Wood