At present, the industry generally follows six basic principles when designing artificial intelligence systems, including responsibility, transparency, fairness, privacy, reliability and controllability, among which responsibility and transparency are the cornerstones of the other four principles.

1. Responsible for:

People are responsible subjects with personal values and emotional likes and dislikes, and the principle of responsibility means that artificial intelligence is the product of people, which inevitably reflects the subjective role of people. People who design and deploy artificial intelligence systems must be responsible for the operation of their systems, and they should be responsible for the results caused by their own decisions and actions.

2. Transparency:

The artificial intelligence system should truthfully record the whole process of making decisions and the decisions made by the system, including the algorithms of data collection, data labeling and use, and ensure the transparency of not only its data and system, but also its mathematical model.

3. Fairness:

Fairness means that for people, facing all people in different regions and at different levels, artificial intelligence systems should treat everyone with a fair, consistent and equal attitude, and cannot exert different influences on groups with similar situations. Artificial intelligence system should provide its processing results to man-machine interface indiscriminately, and make judgments and take actions according to the results.

4. Privacy:

Because artificial intelligence involves data, we should respect privacy and ensure data security, which is the condition to win the necessary trust of the public, and public trust is the premise to make full use of artificial intelligence.

5. Reliable:

Reliability means that the operation of artificial intelligence system should be safe and reliable for users in the process of use, so as to avoid the artificial intelligence system from causing harm in unpredictable circumstances or being maliciously manipulated to carry out harmful behaviors, which puts human beings at uncertain risks.

6. controllable:

The principle of controllability requires that no matter how powerful the function of artificial intelligence system is and how the technology develops, its function and ability should always be under the control of human beings. Artificial intelligence itself will not produce primitive, independent, out of thin air desires and needs. As long as human beings do not maliciously set artificial intelligence as "uncontrollable", the consequences of "uncontrollable" artificial intelligence will not appear.