Facebook shut down its artificial intelligence system overnight. Has AI really surpassed human cognition?

With the rapid development of artificial intelligence technology, whether robots can become independent of humans and build their own civilization like in the movie "The Matrix" has always been a focus of close attention from all walks of life.

Just recently, Facebook suddenly chose to shut down the intelligent conversational robot project that had caused widespread controversy. According to relevant foreign media reports, Facebook made this decision mainly because researchers in the AI ​​research laboratory found that the chatbots under development had gradually broken away from the control of the script and developed and used a new language logic without human intervention.

According to researchers at Facebook's AI research department, the intelligent conversational robot studied in this study is entirely based on text. It is not clear whether a system with words, phrases and sentence structures that are different from human language characteristics has been generated during this process.

However, this is not the first time that an AI has been found using English sentences that deviate from its training.

This comes after Google revealed that its AI system for its translation tool has created its own language, where it translates things into its own language and then back out.

Are the plots in science fiction movies really getting closer to us?

Semantic recognition is still AI's weak spot. Project failure may have nothing to do with program out of control

In fact, some people pointed out that Facebook’s closure of the intelligent conversational robot project was not because AI invented any new language.

According to relevant information, the main content of this AI project is text generation, that is, after providing a small piece of information, let the model generate a piece of text that looks relatively "natural" based on this information.

The continuous occurrence of this "language creation" phenomenon may not be because AI has developed "self-awareness", but more likely because Facebook did not provide enough natural data in the early stages of model training, which led to the AI ​​program being unable to find the rules of natural language through "deep learning".

Artificial intelligence is not breaking through human cognition now

As described by Dr. Selman, a professor of computer science at Cornell University in the United States, the main reason for the rapid development of artificial intelligence in recent years is the advancement of "deep learning" technology.

Through deep learning, artificial intelligence can not only imitate the operating modes of the nervous system in the human brain in some places, but also can very skillfully find certain patterns from it when massive amounts of information are input.

In his opinion, through deep learning, artificial intelligence will be very good at sentence deconstruction, and this ability has indeed enabled Facebook's intelligent conversational robot to produce sentences that conform to human language grammar habits. However, if artificial intelligence is analyzed at the level of semantic understanding, the technical weaknesses of AI at this stage are still obvious.

Perhaps artificial intelligence will surpass human cognition one day, but not now.

At present, although artificial intelligence has made amazing technological progress, in essence, conversational robots like Facebook, which are based on deep learning, are still at the stage of classification, clustering and prediction, such as text and voice recognition, comparison and search for similar items, or sentence reorganization based on relevant data.

It can be seen that currently artificial intelligence can only learn and imitate. Although products such as AlphaGo have appeared on the market, which has heated up the market demand, fundamentally speaking, artificial intelligence is still in a relatively immature stage.

In summary, judging from the current level of development of artificial intelligence, general artificial intelligence that can completely replicate human intelligence such as independent learning and decision-making capabilities is still in theoretical imagination.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.