Processing Power: Improving Our AI Friends

Popular Science just ran an article about the next generation of virtual assistants, such as Siri and Cortana. The idea is that, going forward, virtual assistants might become more useful if they exchange portability for power. Instead of living in our smart phones and tablets, they could live in desktop computers. What does this mean for the development of AI going forward? What does this mean to me, someone who can’t stop crying about robot friends?

There are basically three categories of what I broadly think of as “robots and cousins.” This ranges from Roomba robotic vacuums to Siri to projects like Bina48. The categories are for what they specialize in.

What They Can Do

Robots can vacuum, explore other planets, and carry huge loads. These ones often have simple native intelligences or may simply be remotely controlled. Big Dog acts as a pack animal, for example, with military applications. Their appearances are focused on function, rather than aesthetics. A boost in intelligence for these kids is not an especially well-placed power up.

How They Look

There are some goddamn eerie robotic heads out there. And yes, they tend just to be heads. Bust-style androids are being made who can mimic human emotion with startlingly expressive faces. (Nonhuman mimics do exist, like the companion fur seal pup, but they are far less elaborate.) Though they sometimes fall into the uncanny valley, these bots are the most like what we think of as sci-fi robots: looking and acting like humans with something very different under their skin.

How Well They Think

Ah, now here’s the interesting bit, the “and cousins” part of my statement. Most of the more advanced artificial intelligences aren’t robots at all. They don’t have mobile bodies intended to carry out expression or labor. They’re just an attempt to duplicate our clever brain bits. From chat bots who try their hand at talk therapy to Siri rebuffing her users’ unwanted advances, artificial intelligences are a thing. They’re just not all that bright yet.

So what if we took them out of our phones and gave them more robust processing abilities? Going even farther, what if we gave them dedicated systems, taking them off our computers entirely? What if, say, we start stringing those three categories together? Huge strides are being made in each one, but specialization has meant that none of them really embody that ideal science fictional droid. Maybe that ideal is getting closer.

One last idea, before I stop crying about this for a while: our phones, tablets, and desktops are all networking with each other more and more. Bluetooth connections link more and more of our technology. We have smart phones, smart cars, and smart houses.

So how about an AI with more than one brainĀ and more than one avatar? How about a robotic collective, a shared intelligence spread out across multiple physical forms for maximum physical ability? What would happen if we built someone like that?

Published by Joyce Sully

Joyce Sully believes in magic and dragons and ghosts, but is not convinced her next-door neighbors are real. So she writes stories. Really, what else could she do?

2 replies on “Processing Power: Improving Our AI Friends”

  1. This is really interesting. I don’t think I know enough about networking to leave an intelligent response but I do find robotics abs AI really interesting. I’m especially interested in robots doing jobs that humans can’t physically do. For example, there’s this robot that a trained therapist can use to interact with kids with autism. The idea is that these kids are more easily able to open up to a robot. It is less intimidating because it does not have such an expressive face or voice. So the therapist does some work with the child using the robot, and slowly they work their way up to interacting with the child directky.

    We live in an age when people care more and more about safety. We don’t build amazing landmarks anymore because of safety code violations. While part of me wishes we still had amazing human made constructs, maybe the next best thing is to have human designed landmarks built by robots.

    I’ve seen so many movies about AI gone wrong and how giving them desktops connected by networks will lead to Skynet blowing us up, but people I’ve talked to who actually understand computers don’t seem concerned about that.

    You should see this movie “Robot and Frank” about a man who forms a bond with his companion robot (one that is meant to care for the elderly). It’s a little melodramatic and has a weird ending but otherwise very good.

    If you find any cool robot articles, please post them!

    1. Hey, Johanna!

      Admittedly, I don’t have the technical background either. I’m strictly a very enthusiastic layperson with an unusually large number of what-ifs to throw at the ideas.

      On the subject of robots and autistic people, I have enough opinions to fill a book. Literally: that’s what I’m writing about right now. I’m familiar with therapies used on autistic people, including the idea of robots to model neurotypical social behaviors. The short version: for very personal reasons, I’m not real damn impressed with the idea that autistic people need a robot or anyone else to “train” them to behave neurotypically. On the other hand, having, oh, say, a companion bot who accepts a person as they are, including the stims, unusual methods of expression, and sensory processing difficulties that mark a person as neuroAtypical? Well. Like I said. Book.

      I think a huge number of “robot apocalypse” scenarios can be avoided by not treating robots horribly in the first place. Hence my love of robot friendship stories instead.

      I loved “Robot and Frank!” The friendship that develops between them really charmed me. The ending was heartbreaking for me; I’m still sighing sadly over it.

      I’ll definitely be talking more about robots. In the meantime, consider checking out my “bot feels” tag over on Tumblr. I’ve got a little collection of quotes, art, and articles about robots real and fictional.

Comments are closed.