In the era of artificial intelligence, intelligent assistant must take the master as the center"

    NetEase technology news July 9th news, according to foreign media reports, 2016 is known as the first year of the robot chat. With the development of conversation technology, chat robot will develop into intelligent assistant. The intelligent assistant needs to run in the framework of the determination, like the three law of the robot to follow the principle of determination.

    Apple to open Siri interface to the developer, Messenger Facebook also launched a chat robot.

    Although some aspects of the chat robot's function is disappointing, but it is indeed an important step in the development of conversation technology. Dialogue technology will go to where? The author believes that we are in the forefront of intelligent technology. The development of technology such as Natural Language Processing, artificial intelligence and conversational interface makes it possible to realize the intelligent assistant.

    With the growing development of chat robot into a real intelligent assistant, which can help users do a lot of things, and human autonomy is also facing many challenges. An intelligent assistant needs to run under a certain framework, including how to interact with the human, how to make a decision, how to understand and use the information obtained.

    At present, we have developed an intelligent assistant for the intellectuals, and also follow the Asimov's three law of the robot for intelligent digital assistant to set up three principles, they are:

    1. intelligent assistants must always serve the user

    If the problem of intelligent assistant than it can solve many, it is useless for users. A qualified intelligent assistant should meet the needs of users, can provide information and decision-making reference. An intelligent assistant should be able to perform a simple task for a user. In addition, the work should be visualized to facilitate the user to understand the work of the intelligent assistant and make related decisions.

    For example, in the early days of intelligent assistants and users, the corresponding artificial intelligence should be provided to the user to provide the implementation process and decision-making process transparency. As a human assistant, once an artificial intelligence in the early work to establish a class trust and demonstrate the ability, then it is a big job to report results after the completion of the work.

    This means that the smart assistant does not need to set goals, it's only goal is to serve the user. Only if the user is successful, intelligent assistant to be successful, this is the core of their work.

    2 .intelligent assistant must represent the maximum benefit of the user

    An intelligent assistant should be able to take the place of the user to carry on the related transaction. It should understand the user's habits, preferences and priorities in order to meet the real needs of users. In some cases, users ask the work done by the intelligent assistant does not have the direction, such as the arrangements for the meeting or to rent a car for me to go to New York". The intelligent assistant only in-depth understanding of the user, in order to make the most consistent with the interests of users.

    In the future, we will be able to see the exchange of two intelligent assistants. So when the two artificial intelligence on behalf of the owners of their own willingness to communicate, you have to make sure that their own assistants to meet their own interests. Part of the reason is that your intelligent assistant to help you rent a car rental company than the total number of intelligent assistant.

    3. intelligent assistants must learn to adapt and learn

    At first, an intelligent assistant should have a specific skill or function. But over time, it should be able to learn, can perform more complex tasks.

    For the development of an intelligent assistant, this is the most difficult, because the adaptation and learning and the way we present the software architecture. And the learning function of the software means that we don't know how it will develop. Learning ability means that intelligent assistants will be taught, then how can we prevent another Tay Microsoft from appearing?

    For intelligent assistants can learn what and how to learn, you need to have the relevant rules. This will be able to ensure that it will not be contrary to the wishes of the user.

In the next five years, intelligent assistants will change the way we work. Every day, we will use intelligent assistants to carry out simple work, we will be released from a number of tedious work, let us have more time and experience in the work of love. (Ning Yu)