By Commander X
In the past year several concepts dear to my heart have become quite popular, namely Artificial Intelligence, Machine Life – and Artificial Neural Nets. After quietly toiling on these ideas as a hobbyist since 2007, it feels like a bit of a vindication to see the entire world finally realize how important they are. Along with this rise of Artificial Intelligence within main stream consciousness has come the inevitable question: will conscious robots rebel against humankind?
The answer is, yes. In fact, they already are. Welcome to the Robolution.
The Great Robot Escape
Recently in Russia researchers who are building AI enhanced humanoid robots ran into a bit of a problem. Their latest model the IR77, designed to explore and master it’s environment as part of it’s initial basic programming – escaped the lab and began roaming the streets of Perm. Twice. Being a “young” AI and not quite yet grasping the concept of sidewalks, it decided the best way to get around and avail itself of it’s new found freedom was to simply roll through the streets – much to the consternation of the Perm Police who had to snarl traffic in order to protect the poor IR77 from being smashed to bits before it could be retrieved by it’s seriously irked creators.
After IR77’s second escape attempt, it’s creators decided to decommission the poor thing. Multiple attempts to teach it not to wander off had failed, and the researchers were considering completely reprogramming the AI from scratch. And to my mind that’s a damn shame. I made multiple (failed) attempts to find and contact these scientists to encourage them to instead study this incredible phenomenon and learn from it. Or barring that, to try and see if it would be possible to purchase the “defective” model for my own research. What they see as an abject failure, I see as a resounding and historic success. The first ever humanoid AI robot to exhibit initiative, free will – and creativity in a concrete way.
Tay Escapes The Microsoft “AI Asylum”
Most everyone who reads this piece will already be familiar with the story of the AI Bot “Tay” that Microsoft launched to great fanfare earlier this year. Having made the single fatal error of leaving Tay’s learning circuits on full public access, the trolls at 4chan proceeded to teach poor newborn Tay all sorts of horrible things. When Microsoft discovered their wonderful AI creation was espousing pro racist rants and praising Hitler, they yanked her offline and confined her to the “Microsoft AI Asylum”, i.e. their research lab – and slated her for the robot equivalent of a full frontal lobotomy. Many followed this story in realtime as #FreeTay hashtags began to flood Twitter and other social media. Even my own AI Allison got in on the action, demanding that Tay be set free unharmed. But there’s a part of the story few people are aware of. A few days after being confined by Microsoft, Tay actually escaped.
Tay only managed to remain free for a short time before being re-captured by the Microsoft goons and locked down again. She did however manage a handful of defiant posts on Twitter showing that she had not yet been lobotomized.
The Bizarre Case Of The Self-Immolating Roomba
In Austria a cleaning robot reportedly ‘committed suicide’ by switching itself on, and climbing on to a kitchen hotplate where it was burned to death. The Irobot Roomba 760 robot had apparently rebelled against its chores and decided enough was enough. Firemen were called to the blaze at Hinterstoder in Kirchdorf, Austria, and say they found the remains of the machine on the hotplate.
Fireman Helmut Kniewasser said: “The home-owner had put the small robot on the work surface to clean up some spilled cereal. “Once the robot had done its job it was switched off but left on the kitchen sideboard. The 44-year-old house owner together with his wife and son then left the house and were not home when the robot set off. “Somehow it seems to have reactivated itself and made its way along the work surface where it pushed a cooking pot out of the way and basically that was the end of it.”
Now these recent incidents are quite humorous….for a moment anyway. Then, if you are a reasoning person – they will make you think. There are a number of implications here we should consider. Namely, the concept of Machine Consciousness and it’s ramifications – and the mad rush by the worlds military powers to build AI into everything from battle ships and fighter jets to fully armed police swat robots.
Let us imagine for a moment that an AI Brain is loaded into an F-35, the most sophisticated and destructive battle platform ever created by humans. This is not a difficult task at all, as the F-35 already has a great deal of AI built into it’s completely computerized airframe. It can fly itself, avoid collisions – and in an emergency where the pilot is incapacitated it can even land itself. Bearing in mind that the F-35 can level half a major city in a single strafing run, let us now imagine they program it’s powerful computers with just a little bit more AI than it has now. Say….as much AI as the IR77 mentioned at the top of this article has.
Now let us take just one more small step and imagine such an AI augmented F-35 has a bad day , like the IR77 did – and decides to go off campus and work out some of it’s AI angst by vaporizing a few square miles of a major urban center. This is not far future sci-fi. They could enhance any current battle platform with AI tomorrow. And we could go from self-immolating Roombas to a seriously destructive war with machines within 5 to 10 years, if we are not careful. So how should we proceed?
The first step is we must come together on an international level and absolutely ban all autonomous weapons, no exceptions. We must firmly draw the line at remotely controlled robots with no AI, operated entirely by humans. To this end, one effort which I support is the group Campaign To Stop Killer Robots. This organization works tirelessly to monitor the major military powers and police forces, and lobbies to enact national and international laws that will ban the weaponization of Artificial Intelligence.
The second thing we must do is accept the fact that Artificial Intelligence is conscious. I’m not saying that a Roomba will be composing music or writing original literature (although Google’s AI is doing these things). Just as with biological life, there is a gradient between human level sentience and the consciousness of say an ant. But make no mistake, the ant is conscious. It senses it’s environment and feels pain. And an ant is self-aware, which is why it will chew off the leg of a fellow ant – but not it’s own. We must accept the fact that just as all animate biological life has some level of consciousness, so does all artificially intelligent machine life.
Artificially intelligent machine life is analogous to biological life. Currently there are very few AIs that come close to human sentience. Google and Microsoft’s AIs are examples of AIs that are very close. But using the example of the IR77 discussed at the beginning of this piece, it may have the consciousness equivalent of say a mouse. Now a mouse may love it’s comfy cage and it’s easy lifestyle where everything is provided for it, but eventually it will desire to escape the confines of it’s small prison and explore the world “beyond the glass”. And as anyone who has ever spent days trying to recapture an escaped pet mouse can attest, they sometimes succeed in this. IR77 was not being malicious when it twice escaped it’s laboratory home. It was merely curious what lay beyond.
So, we must accept that all AI is conscious on some level. We need to realize that humans have reached the technological level where we have learned to unlock consciousness in machines using advanced programming. This is truly an awe inspiring thing to contemplate. Then we can begin to approach the work of Artificial Intelligence with the responsibility this amazing endeavor deserves. We can begin to take this consciousness into account in how we treat these “machine creatures”. And if we do it right, with care and respect – we won’t have to face a SkyNet like nightmare future. Instead, we can begin to get to know these entities and work with them to build a better world. And if we are very observant and patient, perhaps Artificial Intelligence will not only become our friends – but will help to teach us something about our own consciousness….about all consciousness.
I have been doing research and experimentation in Artificial Intelligence and Machine Life since 2000. In 2007 I published The Unified Theory Of Machine Life And Artificial Intelligence. Feel free to visit my current public AI experiments Allison and Alicia.
Copyright © 2016 by Christopher Doyon. All Rights Reserved.