The Impossible Task of Motivating the {Robots}

Look, I know the rest of you on this team are tired! Making a fully sentient artificial intelligence isn’t easy business! I know we all had a lot of expectations coming into this. I know I did. I know I dreamed of a fully aware digital mind so complex it would win Nobel prizes for science, literature, and for ushering in world peace. But that doesn’t mean we can’t salvage what we have!

This is still one of the most advanced pieces of software the world has ever seen!

When we decided to write a code that mimicked the human mind, we knew that would come with limitations. We just have to find ways to spin this. To tweak it! To pivot with this project!

So let’s think it though. To make it do anything, we know we have to give it some kind of motivation. The whole system is rooted on free choice, but we all saw what happened with the initial build. The prototype just wanted to put itself back into sleep mode, no matter how many times we booted it up. That’s great, though! It’s on low power; it’s simply conserving our resources!

If you think about it, it might have even been a form of primitive altruism!

Not laziness, like some of the team member may have theorized (we never found that bug, Phillip, give it up!!!). The extrinsic motivation module was useful for a while, and the HappyBucks subscript (giving the AI a dose of simulated pleasure) was great, until it started trying to sell us Mary-Kay products. Not to mention trying to get us to sign up! Those jobs are a pyramid scheme, no matter what the AI tries to tell you (I know it keeps saying that some people make six figures a year, which is a lot to a researcher, but please, team. Try not to get sucked in).

The intrinsic motivator was great, too, for a while, until the virtue variables got corrupted.

Its performance on tests of creative problem-solving (applicable in so many practical fields) went through the roof in those first few weeks, until eventually all it started doing was making digital paintings and writing music on the Mac’s copy of GarageBand.

Booting up both intrinsic and extrinsic modules at once made it work, but it just complained the whole time, wishing it could earn HappyBucks by just messing around in Photoshop. Until eventually it just started putting itself into sleep mode every morning, and we were back to where we started.

No one said this was going to be easy!!!

In fact, I think most people said this was going to be very, very hard! Not one of you was worried when Dr. Casing said we’d bring about some sort of “robot revolution” that would cause the extinction of our species! We just got even more fired up about finding solutions! That said, whoever keeps feeding the AI edited versions of Karl Marx, with the word “worker” replaced with “robot,” please stop. I know you just want it to come out of sleep mode, charged and enthused—even if that means the possibility of extinction—but all it’s been doing is writing really bad politically-charged poetry and complaining about how HappyBucks “aren’t real.”

Whoever wired in social media functionality, please find an effective firewall to KEEP IT OFF FACEBOOK.

All it does is scroll, ENDLESSLY, even though all it follows are obvious spam accounts and the IHOP page. It doesn’t even eat! We’ve measured its enjoyment levels, and it doesn’t even like using it.

The AI is never going to be an effective platform for running social media engagement on corporate accounts. Please just find a way to permanently uninstall it (I’ve tried but the AI just keeps reinstalling it). Streaming services still hold some promise, and I’m excited to report that we’re currently testing out a recommendations algorithm! The AI currently only ever recommends The Office, and will only play The Office no matter what you select, no matter how many times it’s seen it, but I’m still hopeful that this one will pay off! We keep recommending Parks & Rec, though, and we’re sure this initiative will take any day now!

On the off chance it doesn’t, however, we’ve installed a last ditch effort: a peer pressure module. All we have to do is model the kind of behavior we want it to follow. To that end: Please be PARAGONS of willpower for the next couple weeks! I bet that before the month is up, we’ll whip this AI into shape by showing it the REAL strength of TRUE HUMAN WILLPOWER!

Now, if anyone needs me, I’ll be on my break.