When robots in HBO’s Westworld malfunction, they tweak out. The “hosts” stutter, twitch, and generally appear to be suffering from android seizures. Some of them spout lines of Shakespeare, or prophesy doom for their creators. The show is set in an ambiguous but imaginable future, based on current robotic and artificial intelligence achievements, so one might wonder about the accuracy of these robot glitches.
Enter Stephan Bugaj, Vice President of Creative at Hanson Robotics. Bugaj oversees personality design at Hanson, and also works with film and TV partners. If any company is on track to make Westworld-level androids, it’s Hanson Robotics, and if any person at Hanson Robotics is ideal to answer Westworld questions, it’s Bugaj. Plus he’s been watching, and he’s a fan.
Sophia is, in a sense, a prototype of Westworld’s Dolores: Sophia’s physical appearance is crafted in what we can imagine as a lab, as in Westworld, and her personality is crafted by engineers, also as in Westworld. (Rest assured that Hanson Robotics has no intention of creating slave robots like those in Westworld, though.)
So how accurate is what Dolores goes through in episode 2? How about what her “father,” Peter Abernathy, goes through? Or the saloon’s madam, Maeve Millay? Yes, these are just human actors with some special effects, but the show is no less predicting the future. Are they doing it justice? Bugaj fills Inverse in on the state of the art.
Does robot malfunction as seen in Westworld look anything like actual robot malfunction today? Like, if Sophia were to break down, would her breaking down resemble Dolores’s dad’s breakdown?
Right now, robots, when they break down, they kinda just stop working — usually. Sometimes, they’ll start babbling and saying really ridiculous nonsense. Like, truly just strange, inconsequential lines of code or noise or words from their word tree. Or they’ll just break. Motors will burn out, and parts of the face might sag, or arms might stop working. Things might get twitchy, and jittery, or lock up, or just shut down. That’s basically how robots break down right now.
[In the show,] they’re telling an emotional story, and they have characters, and so their breakdowns need to have a sympathetic and empathetic component to them. Whereas right now, machines just break in weird ways.
But the thing is: Could you imagine a robot having a breakdown if it had the characteristics that the machines that they’ve created in their fiction have? Yeah, absolutely. It’s totally believable in the world they’ve created.
What about the whole reciting lines of Shakespeare thing when they break down?
Sure. Why not? That’s actually closer to things that happen now, because we will sometimes get A.I.s that, depending on what kind of internal structures they have for conversation, may fall into a regime where they just start saying things, that might be lines of Shakespeare, that some programmer had given them at some point. And they just happen upon that, and start babbling that out. So that is totally believable.
Has a robot malfunction, or just a robot in general, ever scared the crap out of you?
Not me personally, because I know what’s going on. There hasn’t been a level of transcendence yet where something happens and I’m like, ‘Oh my God’ — like, I either can’t possibly figure out how that might’ve happened, and how did it even — did it change its code, or something like that? — or, ‘Oh, the thing that it did is a sure sign of, not just intelligence, but intelligence going amuck.’ Neither of those things have happened yet.
Bugaj is lucky he doesn’t have to deal with Abernathy — yet. Then again, maybe he never will: Hanson Robotics, again, does not intend to enslave its robots, and so its robots will not be motivated to seek revenge.
This interview has been edited for clarity and brevity.