At present, artificial intelligence has become familiar with tasks such as driving cars, translating languages ​​and recognizing faces. Next, the next goal in the field of artificial intelligence is to let AI have an intuitive sense of reaction to the outside world.
A new article on artificial intelligence published on the Computer Communications Association website found that artificial intelligence has evolved to begin to understand the world as people do without requiring humans to teach the world by carefully entering program data sets.
Yann LeCun, director of research on Facebook artificial intelligence, said that the company and other organizations' research provided artificial intelligence with more opportunities to cultivate human intuition, which is the key to perfecting artificial intelligence. In the next stage, artificial intelligence will be studied through observation, which is similar to that of a newborn baby.
LeCun said to the Association for Computing website: "At the moment, we don't know how to replicate the sense of intuition on the machine. This is a pity. Before we learn how to do this, we will not enter the next stage of artificial intelligence research."
At present, most artificial intelligence adopts the “supervised learning†mode, in which humans use a large number of carefully labeled data sets to train artificial intelligence, so that the latter can complete the recognition of faces or The task of language translation. Predictive learning is a new way of learning in the field of artificial intelligence. In this process, artificial intelligence is completely unsupervised by humans, so there is no need to have a marked data set. Instead, artificial intelligence begins to observe itself and uses its own observations to explore how the world works.
Someone commented on Facebook on the head of Industrial Intelligence Research Yann LeCun: How Tesla CEO Elon Musk has a better vision of the future of artificial intelligence . Musk made a sharp comment on this. In his comments on the personal Facebook news about predictive learning, he replied: "If the only thing you can do is to talk about it, then you can easily become a visionary. But if you are one Self-respecting scientists or engineers, then you can only say so much before you succeed in achieving the flaws you describe."
In essence, it is time for researchers to treat artificial intelligence as a baby and teach it in a way that teaches the baby to see what happens. Facebook's Artificial Intelligence Research Lab is working to make artificial intelligence use intuition vectors to connect the characteristics of things, thereby creating an image of an organic world in its impressions.
At the same time, research laboratories in other institutions have also established neural networks, all competing to create the most realistic images in the artificial intelligence impression. In this way, artificial intelligence can watch the pictures of some movies and predict what the next picture will look like. This is a perception that is closer to intuition, a feeling of the future, not the so-called limitations of artificial intelligence today.
LeCun believes that figuring out how to effectively use this learning method is the way to enter the next stage of artificial intelligence research.
Gas Explosion-Proof Motor,Explosion Proof Motors,Gas Explosion Proof Motors,Explosion Proof Electric Motor
Yizheng Beide Material Co., Ltd. , https://www.beidevendor.com