OpenAI wants to create an omnipotent AI. But in a good way

One Saturday last month, five men aged from 19 to 26 years old surely came out of the cloud of purple smoke in a converted car dealership in San Francisco. They sat for a number of computer keyboards, they were loudly welcomed by the crowd of several hundred people. After ninety minutes of intense mouse clicks smiles fives become dull, and the applause is encouraging. Team OG, world Champions most profitable in the world of competitive Dota 2 games, lost two games in a row the team of bots with artificial intelligence.

This result was remarkable, because sophisticated video games are mathematically more complicated Board games like chess or go. And still, leaning against the wall backstage, Sam— Director of research Institute of OpenAI, who designed these bots to have experienced festive relief.

“We’re all very nervous that morning — I thought that the chances are 60 to 40,” saysHe became CEO OpenAI in March, after the resignation of the President of the influential YCombinator startup incubator. To succeed at a new position,needed the bots that don’t just beat people in a videogame — he wanted bots that are better people around.

The stated mission of OpenAI is to ensure that any future AI is able to surpass “men in the most cost-effective work” will benefit all of humanity. Such technology — General, artificial intelligence, strong AI, or AIS (AGI) — while not even on the horizon looming, but OpenAI claims that she and other rapidly progressing. The organization has shown that it can conduct research at the level of the best in the world. And he earned accusations of inflating the hype on the subject of AI, but also in fixing on the AIS and the potential threats of artificial intelligence.

Planresearch OpenAI — and provocation — should only gather pace. Being first Chairman of the Board of Directors of this organization, he took the post of Director General after he helped translate a large part of the nonprofit staff in the new commercial company, hoping to attract investors with billions required for the formation of the fate of AI and humanity.says that major technological laboratories in Google and the other companies must feel the pressure from colleagues, not seeking to increase its shareholder value. “I don’t want to be in a world where they will create one company that will capture all of its value and will make all the decisions,” he says.

At the MIT event in late 2014, Tesla CEO Elon Musk has described research on the topic of AI as an attempt to “summon a demon”. In the summer of 2015 he talked toand a few others for dinner and proposed the creation of a research laboratory, independent from the technology segment to guide the AI in a positive direction. In the same year was created OpenAI, which turned out to beand Musk. At the beginning of 2018, the Mask left the Board of Directors, citing potential conflicts with the other roles.

During his short life OpenAI positioned itself as a serious place for research AI. Ilya Sutzkever, co-founder of the organization, who resigned from a senior position in AI research at Google, oversees staff, which includes former came from Google, Facebook, Microsoft and Intel. Their work on the topic of robotics and machine learning appeared in the most prestigious conferences. The victory over OG in Dota 2 has brought them respect in the field of AI and gaming.

The transformation of the OpenAI commercial Corporation was caused by the feeling that in order to keep pace with such giants as Google, you will need access to ever-growing computing resources. In 2015 OpenAI stated that she has 1 billion dollars, derived fromMask, co-founder of LinkedIn Reid Hoffman, early Facebook investor Peter Thiel and Amazon. Now,says that one billion will not be enough. “The monetary amount we need for the mission’s success, much more than I thought,” he says.

Applications IRS show that in 2017, when OpenAI showed his first Bota-daughter, she spent 8 million dollars on cloud computing. Since then its costs are likely to be considerably increased. In 2018 OpenAI announced that the predecessor system, winning OG, tied more than 120,000 processors leased from the cloud Google within weeks. Winning the Champions version have been training for 10 months, equivalent to 45 000 years of playing Dota 2 against itself. How much did it cost? Greg Brockman, CTO OpenAI says the project took “millions of dollars”, but did not disclose details.

Altman is not sure that OpenAI will continue to rely on cloud services competitors — it is still open to purchase or even to develop hardware for artificial intelligence. The organization is closely monitoring the new chips, developed by Google, and many start-UPS to squeeze out of the algorithms of machine learning even more.

To raise funds necessary for access to future hardware,tried to sell the idea to investors under the scheme, wild even for Silicon valley. Put the money in the OpenAI, wait, and the company will return your investment a hundred times — as soon as we invent robots that surpass humans in most economically valuable tasks.

Altman said that expectations were “the most interesting experience of fundraising in my life — it is not suitable for any other model.” The biggest interest was, in his words, to show interested in AI rich. Hoffman and venture capital firm Khosla Ventures has invested in a new commercial OpenAI, but did not respond to requests for comment. Nobody says when to expect the refund, but bet on OpenAI is not for the impatient. Companies report that the return of funds may exceed the standard in the industry (ten years). “We immediately tell them that you will not receive funds in the next 10 years,” saysall the efforts, OpenAI is often criticized by the industry leaders. In February, the company revealed details about the processing language, which was able to generate surprisingly lively text. Some immediately called the invention “dangerous” and expressed fears that it could be misused to litter the social network.

Some prominent people in the field of research in AI was annoyed, including the chief of AI in Facebook Jan Lacuna. In their posts on Facebook, he defended the open publication of research on AI and joked that people should stop having children because one day they will publish fake news.

For some, the episode stressed that the mission OpenAI not much different from other corporate or academic laboratories. “They carried out plus-minus the same study, like others, but want to attract billions of dollars,” said Zachary Lipton, Professor of machine learning at Carnegie-Mellon. “The only way to do this is a little half-hearted”.

Altman recognizes that OpenAI might have sounded the alarm too soon, but says it’s better than late. “The technology industry has not done a very good job trying to take the initiative to abuse,” he says. Self-censorship is necessary and most OpenAI.

After the defeat of OG last month, in order to Express sympathy, OpenAI decided to conduct a second experiment, designed to demonstrate a favorable direction superhuman AI. Experts in the field of Dota 2 and some of the recruits played in teams with bots.

Software, who was unlucky enough to partner with noobs on the team, for the most part showed superhuman indifference to help the beginning player. Instead, it focused on winning instincts, honed by months of expensive training.

No one knows how to make a small AI more flexible or what components may need to make the AIS more than the technological fantasy. Evenfrightened by the magnitude of the problem. “I have days when I believe everything that happens, and days when it all seems a distant dream,” he says.