He can read lips and create new recipes for food. He can beat anyone in chess and go. Every major technology company is trying to integrate it into its organization and business. And, it seems, every new application feels obligated to say that it uses machine learning to some extent to make your life better. Artificial intelligence can be called without exaggeration the most discussed topic of the outgoing year. The revolution is close, and the most obvious question that we must ask ourselves is: what's next?
This question was asked to answer James Hendler, director of the Rensselaer Institute for Data and Applications Research, one of the developers of the semantic web. The problem, according to Hendler, is that many people see in the AI such a "Terminator" or a utopian dream, completely excluding humanity from the equation.
"People want to paint this technology with white and black colors," he explains. "But best of all people cope with the shades of gray."
And before turning to reasoning about the future of artificial intelligence, Hendler gives out a slightly political slogan: we humans and AI are strong together.
AI package for mass programming
"I think that in the short term I'm most impressed with AI in the way it's just used by programmers. This is no longer the lot of professionals, "says Hendler.
The classes that he currently teaches on the subject of AI's cognitive calculations illustrate this point of view. The magistrates make projects like chat bots that can answer questions about the Harry Potter universe in just a few weeks. A few years ago such a feat would be the topic for the Ph.D. thesis.
What changed?
You no longer need to build in-depth training, computer vision or natural language components from scratch. Just download the package with open source and integrate it into your system, slightly changing. This is similar to the game with WordPress, although Hendler prefers to talk about the days of the birth of the Internet. In the early 1990s, with a little understanding of HTML, it was possible to create a website, by installing a kind of pre-packaged code.
"Artificial intelligence was packaged in a usable way," says Hendler. "It's enough to collect it in pieces and make it work without conducting fundamental research on what these components are."
Open doors for innovation
In the short term, says Hendler, this opens the game for players of all stripes.
"We are seeing a lot of innovation in small companies that use existing methods for in-depth training, sight and language processing," he says. "Heavyweights - Microsoft, Google, Facebook - will invest heavily in this technology, but already in new directions."
Meanwhile, in scientific and government circles continue to study the evolution of technology-related AI. Hendler cites as an example self-managed transport, first developed by universities like Stanford to win the DARPA Grand Challenge. Then Google finalized this technology. Today, it seems, every automotive company on the planet is engaged in the withdrawal of self-governing cars on the road.
While there remains a need to develop new artificial intelligence technologies to solve problems, Hendler says that in the short term, the focus will be on various business cases that can be done with the use of existing tools.
"I think that such innovations attract entrepreneurs and start-ups. There will be a lot of them, "says Hendler.
Solving the problems of developed and developing countries
What can an ordinary technology user get from AI in 2017 and after? In this case, "more" means "less", because this technology works invisibly in the background. It will not be obvious, that it could be bought and see the whole world in a different light.
Take, for example, Siri, the ubiquitous virtual assistant from Apple. Siri's competence in performing more complex tasks is constantly improving, but it still (and often) by default refers to a search on the Internet. But one day, in the near future, someone will ask Siri or her counterpart to "show a picture of children from lunch today" - and the machine will quickly and accurately give the result.
In fact, a number of start-ups can already boast of such achievements. Snips uses an AI method called "context aware" to create a kind of memory on the user's mobile device, sorting through contacts, mail, calendar, photos, and so on. The system learns what is important for the user in his life, acting as a single portal for all applications and information stored on the device.
"The thing here is that artificial intelligence makes technology disappear, so that they do not bother you anymore," says Rand Hindi, CEO and founder of Snips.
Of course, this is all a problem for developed countries - for technologies to disappear. But Hendler believes that in the near future projects will be developed that will lead to better conditions in developing countries. In particular, he and others are working with IBM to provide literacy to a billion people in the next five years.
"Perhaps we will be able to significantly change the lives of a huge number of people, especially in those countries where the literacy rate is currently low. This is where people suddenly see how technology is changing their lives, like never before. "
Inevitably, together with the development of artificial intelligence, our world is waiting for coups and, maybe, even strikes. For example, truck drivers who may be left without work due to the spread of self-governing vehicles. But at the same time there will be a lot of jobs and, probably, the same people will be able to receive income not lower, but maybe higher than the previous level.
The article is based on materials .
Comments
Post a Comment