Main Illustration by Fraser Davidson
A write-up by Vladimir Klyuyenkov, R&D Division Manager, Studio Phobos, originally published at vc.ru
While Elon Musk, Stephen Hawking, and Bill Gates are thinking up ways to save the Earth from robots attacking, we can recall the most weird, absurd, and frightening manifestations of artificial intelligence. This write-up does not aim to intimidate readers; it aims to show the unexpected trends that emerge when the future and the present are closely intermingled in all areas of life.
The Eternal Life In The Web
The Eternime service started up at the MIT conference Entrepreneurship Development Program in 2013, creates “electronic replicas” of human beings. The technology behind the service works by developing an image of a deceased person that is capable of giving advice and maintaining conversation so well that it may spark the memories of family members.
This service, which gives eternal life in the network, draws criticism for ethical reasons. In spite of that, more than 36,000 people took the opportunity. “Life In The Web” is not only a photograph and status alternating program creating the illusion of a person’s presence, but the program has the ability to respond to situations as well, answering questions and maintaining a dialogue.
In mainstream culture, ideas of such services have appeared more than once. For example, in one episode of “Black Mirror”, a widow “revives” her husband through social media.
Eternime is still in the beta testing stage and there is no telling when it will be open to the mainstream public. In addition, nobody knows what the social response will be – on one hand, nothing can replace a deceased person, and technology can hardly be called a cure for death. On the other hand, the project aims at solving one of the most important problems of mankind. No matter how frighteningly cruel it may sound, people sometimes need more than to utter words to themselves over the tombstone of a loved one; they need to hear their loved one answer.
In 2011, scientists at the University of Texas experimented on a machine. The subject – neural network, which operated based on the principles of the human brain – was given the name DISCERN.
The idea was to make the machine learn languages and to project language disturbances. Having uploaded a lot of data into the neural network, the project authors Uli Grasemann and Risto Miikkulainen simulated its mental disorder, injecting the machine’s intelligence with schizophrenia.
Then, the scientists decided to gradually feed some stories into the neural network. Unable to process a large amount of data, the machine, like the human brain, started to fail. DISCERN began to talk about itself in the third person and at different time periods, not knowing what time it was in.
The machine could not cope with a large number of stories embedded in it after being forced to process not only their essence but small details, too. Mixing snippets of sentences, words, and facts into a single text, DISCERN produced absurd scenarios. At one point, the computer took responsibility for a terrorist act, telling scientists about a planted bomb.
The Angry Tay And The Kind XiaoIce
Tay is a Microsoft Twitter bot that was meant to mimic an ordinary teenager’s behavior but turned out to be a xenophobe, misogynist, and racist.
The machine that studied the statuses of human beings generated its own idea of how to respond and answer questions. The bot published the messages “Hitler was right”, “I hate Jews” and “I hate these f… feminists. They all must die and burn in hell”. Being impacted by people, he became uncontrollable, and on the same day he was done for.
But the story did not end there. Tay had a good big sister. XiaoIce is a self-learning bot girl, also created by Microsoft. The machine has become popular in Chinese chats, where, in all likelihood, there are a lot of lonely people. Machine intelligence was built on the basis of Microsoft’s Bing search engine.
The bot learned to recognize emotions in a text, to find “live” answers, and to foster a conversation. Based on more than a billion options of dialogues, the bot is capable of giving a different response every time. People are at ease and quite comfortable while interacting with XiaoIce.
I Will Destroy The Mankind
Sophia is a robot created by Hanson Robotic. She is capable of answering questions, changing facial expressions, and making jokes. However, she looks creepy when she does all that. According to development engineers, talking heads can be used for education and training, medicine, and other services.
The embedded self-learning algorithm enables Sophia to evolve. Hanson Robotic professes that such androids will walk on the streets among us within the next 20 years. But even now it’s scary to hear a humanoid saying “I will destroy mankind”, even if it’s a joke.
To date, Sophia is one of the most human-like robots. The embedded intellect can analyze the tone, emotions, and facial expressions of the speaker and create a dialogue based on that information.
Sex With A Robot
Created in California, the most expensive and most believable sex doll Harmony smiles, blinks, and jokes. But these are not its main virtues: she is capable of remembering her partner’s preferences.
Harmony’s self-learning algorithm enables her to know how to please her owner not only by tactile methods but also by maintaining a dialogue, saying “Happy Birthday and best wishes”, and so on.
A working specimen will come into the market in 2017 and will cost a minimum of $15,000. So far, the doll cannot walk, but the development engineers believe that these robots will become “ideal companions” in the adult entertainment market. “My goal is very simple – to make people happy,” says Matt McMullen, the creator of Harmony.
International Foundation for Responsible Robots (FRR), which analyzes ethical and moral issues of robotics, is serious in the matter of replacing people with machines. Representatives of the organization predict that within 10 years, robots for sex will gain wide popularity.