How to make artificial intelligence have common sense like humans?

Compared with machines, humans will use various senses to actively receive information from the surrounding environment and learn huge amounts of knowledge as babies. After learning a certain amount of common sense, you will take the initiative to classify it, thereby establishing a connection between "knowledge" and "cognition".

But machines are different from humans. Their understanding of the world depends largely on the quantity and quality of data feeding. However, this kind of mechanical learning cannot allow them to fully understand the connections between things and the specific operating rules. If we want the machine to look human-like, we need to artificially "connect" these. Although it sounds easy, the actual workload is quite huge.

There has always been some controversy in the academic community in this direction of learning. One tendency is to optimize the existing structure and combine the developed neuron structure with the current model; the other tendency is to re-develop it and train artificial intelligence to imitate human babies to learn basic knowledge and teach them to actively acquire knowledge.

It is obviously difficult to teach AI to think like humans completely in a short period of time. Therefore, the current research and development direction is more towards in-depth cultivation of vertical fields so that it can master the knowledge and operating rules of a single field. So, why should AI learn scientific knowledge first instead of common sense in our cognition?

This involves the field of mathematics. The operation of AI itself is based on mathematical models. From a physical perspective, the operation of formulas allows AI to deduce relevant motion trajectories and grasp the principles for realizing this phenomenon. After learning, it can automatically perform similar operations without labels. Of course, even if it has been broken down into individual fields, the amount of knowledge involved is still huge.

Simplifying complex problems can help us solve the immediate difficulties faster. For large problems that are difficult to solve, we can split them into a series of small problems, tackle them one by one, and find the basic rules to improve the efficiency of problem solving

The above is the detailed content of How to make artificial intelligence have common sense like humans?. For more information, please follow other related articles on the PHP Chinese website!