The Artificial Intelligence Behind Bender

Hey, everybody, I’ve got a new series. I’ll review fictional robots and offer real-life AI techniques that might be able to reproduce some of their famous behaviors. Since fictional AI tend to be highly advanced, they often require many different techniques, so I’m going to focus on explaining one system at a time in each post.

Starting this list off with a bang, I’m starting with the ultimate “AI Behind.” One of the shiny metal variety, to be precise. It’s everyone’s favorite chain-smoking functioning alcoholic industrial construction robot, Bender Bending Rodríguez. 

As a robot, Bender likely relies on some form of reinforcement learning. Reinforcement learning is a field of artificial intelligence based on selecting actions in an environment to maximize predicted rewards. Reinforcement learning comes from the science of animal behavior. Like a dog learns a sequence of actions that is likely to earn it a treat, so does a reinforcement learning system learn a sequence of actions likely to earn it a reward. Bender does not receive direct feedback for each action he takes, so being able to imagine what might give him a reward down the line is an important skill, especially when he’s hatching an outrageous scheme.

But hold on, you might think that Bender is impossible. Who would program an industrial robot to be a foul-mouthed, amoral jerk? Enter Microsoft.

Microsoft Tay was an AI released online by Microsoft in 2016 and shut down sixteen hours later. In fewer than 24 hours, the cesspool that is Twitter had transformed it into an openly racist, sexist, transphobic holocaust denier. It didn’t run on alcohol or smoke cigars and it had no shiny metal derriere to invite people to bite, but it was proof that artificial intelligence does not have to be couth:

Uh… wow. I think I’ll stick with Mr. Rodriguez, thank you.

Name: Bender Bending Rodríguez

Origin: Futurama

Likely Architecture: Reinforcement Learning, Convolutional Neural Networks for vision processing, and Transformers for speech and language. Neurosymbolics for simple reasoning.

Possible Training Domains: Twitter, 4-chan, Youtube Comment Section

I take requests. If you have a fictional AI and wonder how it could work, or any other topic you’d like to see me cover, mention it in the comments or on my Facebook page.

By Sam Munk

Science fiction and Fantasy author with a focus on philosophical inquiry and character-driven drama.

1 comment

  1. Hey Sam, great read! I particularly enjoyed your in-depth discussion of reinforcement learning, since it was something I hadn’t really thought of before. Being a fellow blogger myself, I also really appreciate how organized and well-formatted everything was – it definitely made the content much more digestible overall. Keep up the awesome work!

Leave a comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: