Google is developing a “red button” to disable AI in the event of a threat

Worried about the development of the artificial intelligence that he created, Google is developing a “button” that can turn off the AI ​​in case of danger and threat to a person. This news, replicated by journalists in June 2016, added fuel to the fire of concern for the future relationship of robots and people.

In authoritative Western publications in 2016, information appeared that Google was developing a “big red button” designed to stop AI in the event that artificial intelligence threatens humanity. According to the information of journalists, for this purpose the specialists of the Google division – Deep Mind – teamed up with scientists from Oxford University. The study allocated 400 million British pounds. “If the agent (AI) is running in real time under human control, the operator needs a large red button, which can be clicked, preventing unwanted actions on the part of the agent,” the online publications explain.

Previously, indeed, a document entitled “Safely Interruptible Agents” was posted on the Web. Its authors are Lauren Orso from Deep Mind and Stuart Armstrong from Oxford University. The document describes a possible future neighborhood of people with artificial intelligence, opposing the desires and aspirations of mankind like fantastic Skynet and Terminator. This does not seem insane considering that not so long ago the oldest and largest American law firm, Baker & Hostetler, “hired” a robot-lawyer to assist in the processes. Artificial intelligence is supported by IBM Watson supercomputer. As for Deep Mind, previously the real hero of the news was Alpha Go, an artificial intelligence developed by the company, who beat one of the strongest players of the go, grandmaster Lee Sedol. This victory was the reason for the sharp forcing by South Korea to develop its own artificial intelligence and allocate significant funds for this from the budget. It should be added that the head of Deep Mind Demis Hassabis urged people to be cautious about the information released. In his Twitter account, he noted that, despite Internet rumors rumors, the company does not yet forecast the future of the artificial intelligence Alpha Go. “How to decide, we will make an official statement,” the head of the company explained.

Champion of Europe Fan Hui for playing in go / © wikipedia
Elon Mask: “We are already cyborgs”

Anxiety about the upcoming complications of the relationship between robots and people is growing. And this is the main adepts of scientific and technical progress talking about this alarm. For example, Elon Musk, who stated in August 2014 that the technology of artificial intelligence is “more dangerous than nuclear weapons.” And a few months later, speaking to students at the Massachusetts Institute of Technology, the creator of Tesla and SpaceX called AI “the greatest threat to our existence.” At the same time, by the end of 2015, Mask himself became a co-founder of the non-profit research company OpenAI, explaining its goal by the need for openness of the process of creating artificial intelligence. “If you have a” button “that can do bad things in the world, you do not want to share it,” Mask explained.

At the end of May 2016, speaking at the Recode annual conference, Elon Musk told about a certain project, or maybe just the idea of ​​”neural lace”, which should reduce the risk of mankind to become a favorite toy in the iron hands of robots. “For example, you have a limbic system, the cortex of the brain, and this is the third system – a digital layer that is well compatible with your body. How the brain works in symbiosis with the limbic system, so this digital layer will work seamlessly with the rest of the body, “said Mask. This neural system should increase the level of reception and transmission of information by a person, the ability to process it. “We are already cyborgs,” Mask said, explaining that the Internet, mobile phones and other gadgets, in fact, have developed in humans superpowers. “But we are limited in the reception and transmission of information. Especially in the program, “- added Mask. People are able to receive and interpret a large amount of information, but it is much more difficult to generate this information. A person can only speak quickly, write quickly or print. “Neural lace”, connecting a person to a digital communication system, could overcome this biological limitation. Mask did not begin to explain whether this “lace” would be implanted into the body or grown naturally, say, with the help of gene editing, but noted that it can pass “through the veins and arteries”, providing paths for the neurons.


© Depositphotos

However, there is an opposite point of view. Chairman of the Board of Directors of Alphabet (Google) Eric Schmidt in an interview with Tech Insider was skeptical about the performances of both Elon Mask and his colleagues Stephen Hawking, who had also warned about the possible hegemony of AI. “Stevan Hawking is certainly a brilliant man, but he is not an expert in the computer field. Elon is also a wonderful person, but also, more likely, a physicist than a computer scientist, “Schmidt said. On the potential future scenario in which robots destroy humanity, he stated: “This scenario simply describes an option in which computers become so smart that at some point, for example, the bug in the program decides to destroy humanity. My question to you is this: do not you think that people will notice this and start turning off the computers? We will have a race between people turning off computers, and AI, frantically trying to move into a device that has not yet been turned off. And so to the last computer. And we can not turn it off. It’s a movie, a movie. The existing state of things in the world does not support such a scenario. ”

Scalable robotics
Co-chairman of the existing at the NASA University singularity Neil Jacobstein told the magazine PCmag that for the period from 2011 to 2015, the world’s companies have invested in the study of AI more than three billion dollars. In addition to large investors, including Facebook and Google, there are also a number of lesser known start-ups. For example, Brighterion uses artificial intelligence in cloud analysis and industrial IoT, Feedzai trains robots to detect fraud in banking operations, Experfy with AI effectively solves business problems of merchants and so on. That is, the scale of the development of artificial intelligence in the world is now not limited to a dozen of the main players of the IT market – it is much larger. In his interview, Jakobstein, in fact, continued the idea of ​​Mask, also insisting on the need for the development of the human brain to resist the synthetic. “We all share the same problems. The brain is limited in speed, memory, bandwidth and constrained by prejudices, “Neil Jacobstein said.

According to the scientist, the trend in the development of AI is now moving towards the creation of an artificial neocortex (a new cortex that makes up most of the cortex in humans, in contrast to the lower mammals) and the hippocampus. Thus, the robot’s brain will be as human, only many times more powerful and faster. “AI is becoming more powerful, and we need to decide how we will behave in the framework of the ongoing co-evolution. Both in terms of the tasks that we put before artificial intelligence, and in terms of how we want to build our relationships without losing human self-esteem, “Yakobshtein explained. According to the scientist, initially the interaction of people and robots will be an “elitist process”, but then it will enter everyday life, as it used to be with cellular phones. “Through your smartphone, your brain gets access to all the world’s knowledge and powerful artificial intelligence that processes digital information. You already have a symbiosis with AI, “concluded Jakobstein.
Flesh of the flesh
However, it may be that the threat to mankind will come not from a set of glands and wires, but from organic “flesh from the flesh”. At the end of May 2016, the world scientific community was intrigued by the last “secret meeting at Harvard,” whose participants want to create an artificial human genome and, accordingly, the man himself from scratch. The details of the meeting on the pages of Live Science were told by a graduate of Harvard University, a molecular biologist specializing in gene editing, Jeff Bessen. As it turned out, it’s not about any Frankenstein child. More than 130 scientists, industrialists and ethical experts gathered at Harvard to discuss the project of creating the artificial human genome Human Genome Project (HGP). The title, in fact, a reference to a similar project, which as a result of which in 2003 it was possible to completely sequester the human genome. That project began in 1990 under the leadership of James Watson under the auspices of the US National Health Organization, and its goal was to determine the sequence of nucleotides that make up DNA and identify 20-25 thousand genes in the human genome.

The creation of the genome has its own history. In 2010, American biologist John Craig Venter deciphered the first genome of the microorganism – the bacteria Haemophilus influenzae – and announced the creation of an artificial cell. Venter introduced the world’s first artificial organism, a synthetic bacterium called Cynthia. Since then, scientists have systematically attempted to advance further in research, including the organizer of the Harvard meeting – the American geneticist Jeff Bouke, who first created a functional eukaryotic chromosome contained in yeast.

However, such work is not cheap. The creation of a synthetic bacterium for the same Venter cost over $ 40 million and took years of work. Experts believe that in the current conditions the production of one human genome will cost about $ 90 million. However, if the case “put on stream” and apply the technology, by 2036 this amount could be reduced to 100 thousand dollars. In addition, there is a production problem – the capacity of all laboratories in the world, according to Jeff Bessen, today is not enough to create one human genome.
From study to creation
As Jeff Bessen points out, at the Harvard meeting, scientists called for moving from reading the genome to building it from building blocks of DNA. According to the idea of ​​the ideological inspirers of HGP, including geneticist George Church, his American colleague Jeff Buke and biotechnology specialist Autodesk lawyer Nancy Kelly, one can “completely synthesize the human genome in the cell line.” The process of creating artificial DNA, according to Jeff Bessen, is analogous to the process of offset printing – the main thing is to collect everything in the right order. The result is chemically identical to the natural macromolecule. The world market of synthesized DNA (we are talking about the existing creation of separate elements of DNA, not a complete genome) is estimated now at about a billion dollars a year. The artificial macromolecule has practical application: it is cultivation of human organs for transplantation, creation of immunity to viruses, oncological diseases, creation of new medicines and so on. Participants in the Harvard meeting expressed the hope that their work would help in “developing safer, less expensive and more effective therapy.”

The secret halo around this meeting is explained simply: the project is quite complex in the ethical sense, and it is not known how the ideas of scientists will be perceived by the general public. In particular, now the organizers actually vetoed any comments regarding the meeting before the official publication of the results of the meeting.

There are many questions to HGP. “How good will it be to study and then create the genome of Einstein?” – a bioprocessor from Stanford, Drew Andy, is puzzled. Dr. Church and bioengineer Andy, by the way, have been cooperating for a long time together – they together created Gen9 company, whose specialization is clear from the name. Some experts believe that participation in the meeting of businessmen and the alleged private financing of the project will result in “privatization of the technology of genetic modification”. It was also suggested that researchers harbor clandestine plans for cloning people with synthetic DNA. According to Jeff Bessen, the chemical production of the human genome – the set of genetic instructions contained in each cell – can indeed give a new meaning to the term “baby from a test tube”. “If such a technology existed, the genome of any person could be deciphered, and then synthesized at the request of someone,” warns Bessen.

Clones from the future
The fear is not really groundless. Chinese scientists, who submitted a project of a “clone factory” at the end of 2015, later announced their readiness to clone people. The world’s largest cloning company, the Chinese company Boyalife Group, will build in cooperation with the South Korean Sooam Biotech Research Foundation and the Chinese Academy of Sciences in North China’s Tianjin. The volume of production of cloned cows, according to scientists, by 2020 will be one million a year. In the future, the “clone factory” also intends to manufacture horses for horse racing and police dogs with specialized abilities (increased sense of smell and search). Another area of ​​interest for the enterprise is the cloning of primates for medical experiments. Therefore, the statement of the leadership of the future factory about the readiness to move from primates to people became logical. “Technology is already there. If this is officially resolved, I do not think that someone will cope better with this than Boyalife, “the executive director of the future factory, Xu Xiaochun, told reporters. He stressed that so far no experiments in the field of human cloning have been carried out by the enterprise.

And yet, despite the need to address ethical issues, according to Bessen, the creation of the human genome still remains a pipe dream. “To date, there is no effective way to transplant an artificial genome into a human cell. Even the most impressive achievements in this regard, such as Dr. Buke’s “yeast” project, are hundreds of times smaller in scale than the current challenge. Making the genome of a fly-bird or nematode fly is ten times simpler and safer in terms of ethics, can answer the same questions that people want to get by synthesizing the human genome, “the molecular biologist believes.

Harvard’s “Last Supper”, he said, has exposed yet another problem of modern science – its conservatism. Journalists learned about the project on the Internet, in social networks. And the official information has a long way to go. “Once, my colleagues and I were waiting for the publication of the necessary scientific publication for six months,” adds Bessen. According to the scientist, his colleagues now need to find a balance between the control of publications and the right of journalists to carry out their work, which ultimately affects scientific information.