An AI startup is building an AI that can predict the future based on the things it has seen, a team at MIT says.
The company, called Neural Networks for AI, is developing algorithms to recognize the characteristics of people in the future, the researchers say.
For example, it might be able to learn more about how people react to specific situations.
“The first step to a machine-learning algorithm is understanding what that is, and the more it understands that, the better it gets at that,” the researchers wrote in a paper published in the journal Science.
The researchers are calling their system the Neural Network for AI.
They say they’re looking to “develop neural networks for the future” and that their goal is to have a working system by 2021.
Their goal is “to develop neural networks that can be used to build applications in real-time that can help humanity better understand the future,” they wrote.
“We’re interested in learning what we can about our own future, but we also want to be able, if we want to, to take that knowledge and apply it to the real world.”
The research team is working with researchers at the University of Pennsylvania and the University at Buffalo to get the AI system up and running.
They’re hoping to create a network that can make predictions on its own and make predictions that are accurate enough to help us better understand our own world.
“It is a very interesting field,” MIT Professor Peter Diamandis told NBC News.
“It’s an area of deep research and it’s not just about AI.
There’s a lot of interesting stuff happening with AI and the ability to predict the futures is something that is important for humanity.
So this is an area where we’re seeing some promising progress.”
The goal of the project is to get a working AI system capable of recognizing and predicting human emotions and actions, and then making predictions about how that will affect the world, Diamas said.
The AI could also use this knowledge to make predictions about what the future will bring, he said.
For instance, the team hopes the AI will be able “to predict the level of interest in certain kinds of content, like news and movies, or maybe the rate at which new technology is making itself available, and use that information to better predict how it will evolve.”
To understand what’s being predicted, the Neural Networks project is working to create “fuzzy models,” the team says.
They want the neural network to predict what is happening in the world “based on what the current information is saying,” the paper says.
That’s a problem, because a neural network can’t accurately predict the world in advance.
If it does, then it would get wrong information.
The goal of this project is “try to build a network of these fuzzy models,” Diamands said.
The team is currently building an artificial intelligence called Atsushi Nakazawa that “learns” to make these predictions based on its knowledge of human emotions.
It’s built in part with data from a variety of sources, including a massive amount of research into what’s going on in the past.
“We have very extensive data that’s available, in the form of many different people who have written very detailed and well-documented work on human emotions,” Dias said.
“And they all have very high levels of confidence that they know what the emotions are, and that they can predict them.”
The team hopes that by developing neural networks, it can “improve our understanding of emotions,” the authors wrote.
They say the system will be “more useful for the human world” and will “help us better predict the human future.”