How will robotic personalities evolve in the future?
Robots that have the ability to think and make their own choices is a fear that has evolved thanks to science fiction books and movies, but how exactly can a robot make those decisions? Within a group of people, each person will make differing choices when faced with similar obstacles and stimuli based on previous experiences and a unique personality. Faced with a similar choice, it is usually assumed that a robot would make a decision based purely on the programming it houses and therefore leave the robot devoid of a personality. This is not going to be the case any longer, because scientists have developed the ability to give robots a unique and changing personality based on robotic genes and chromosomes.
Your future puppy overlord.
A joint venture between Samsung and the Korea Advanced Institute of Science and Technology has spawned Rity (seen on the right), a computerized puppy that has virtual DNA. Rity has an artificial genome made of 14 chromosomes and 1,764 genes. The 14 chromosomes control 14 possible combinations of internal states within 3 units – motivation, homeostasis and emotion. Each chromosome is made up of 126 genes that are able to mutate from generation to generation to give each version of Rity its own unique personality.
To test the artificial genome’s ability to change over multiple generations, the researchers ran a test that spanned numerous generations:
To demonstrate an artificial genome, the researchers used their evolutionary algorithm to generate two contrasting personalities for Rity – agreeable and antagonistic – and compare Rity’s behavior in the different cases. Running the algorithm through 3,000 generations took about 12 hours to generate a genome encoding a desired personality by a Pentium 4, 2 GHz processor. For comparison, the researchers also used manual and random processes to generate genomes with agreeable and antagonistic personalities, though neither outperformed the evolutionary algorithm in terms of personality consistency and similarity to desired personality.
Based on this research, an evolutionary approach to robotic personalities outperformed both manual configurations (programming a robot to respond in a specific way) and random configurations (hoping random responses would be correct). This means that by mimicking human evolution at a phenomenally faster pace – 3,000 generations in 12 hours – a robot can react to external forces more accurately than it could with standard programming.
Since a manual configuration is an attempt to make a robot respond in a way similar to a human, and it doesn’t work as well as artificial genetic evolution, this sure seems like the beginning of robots becoming able to dominate humans in pretty much everything. While this research is aimed at improving relationships between humans and robots, the fact that a robot could evolve through thousands of generations while you sleep is very intimidating. Rather than being afraid of a robot encountering a glitch that causes it to harm humans, the fear could shift to a robot simply doing what it is supposed to, only with unintended consequences.
This research leads to numerous questions of human safety and dominance over machines. Since genetic mutations led to humans developing into what we are today, how would a mutation in artificial genomes alter a robot that can evolve through the same number of total human generations in a couple of days? Would scientists be able to insert a failsafe into artificial genomes to ensure robots would not revolt? How will the interaction between a robot with an artificial genome and a human alter the robot’s evolution?
The answers to these questions do not do anything to calm my fears over this research. With artificial evolution happening at such mind boggling speed, it does not seem that anyone would be able to control how a robot changes from one generation to the next. A robot might interpret external stimuli much differently than a human would, so any expected changes to a robot’s personality could not be estimated with confidence. A failsafe doesn’t seem possible when a robot would be able to adapt its reactions to it so quickly. The intricacies of human personalities interacting with a robot would seemingly cause a robot to mutate in completely unforeseen ways.
Seemingly any answer to a question raised by artificially evolving robots only leads to more questions. All I know for sure is that this is truly terrifying technology.
::from PhysOrg via The Daily Galaxy::
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_e.png?x-id=9cae4f26-aff5-438f-bb3f-77d60cde8938)
More From Robot Armageddon
- Sex With Robots Is On The Horizon and It Could Doom Us All
- Robots That Can Plug Themselves In To Recharge
- The Robots Are Watching You and They Will Remember You
Twitter Username: decembersun
says:
Evolution is always played up as a positive thing, but there are negative aspects to this theory (like enthropy, bad mutations, and systems that break down with time). I wonder how this would apply to the robot world as well? Bad mutations, for instance, along the lines of thinking that “humans are evil.”
Twitter Username: RobotArmageddon
says:
That would be a really bad mutation. There is also the “Crushing human heads is the hot new robot sport” mutation.
Twitter Username: darms_88
says:
Wow.. awesome, I just know robots also have DNA and chromosomes now
Unbelievable, lets just hope they’re not as smart as their creator. Or else they will lose control
Twitter Username: ahardrain
says:
David, Tis the season to be TIRED. I am sorry to have been away for awhile, but the holiday season tends to take all your free time and blogging time away from you. With that being said, I wish you a Very Merry Holiday and hope you get to enjoy many warm smiles with your whole family, kindest regards Glenn and all your friends at MOB
Thank you for advertising on our site also, you been a great blog friend and wish you a great New years,
Twitter Username: ahardrain
says:
Wishing you, your family and all your readers a very Happy Holiday from your friends at Man Over Board
[...] have previously written about robotic software that used artificial DNA to run evolution scenarios on a software dog [...]