In Why Does The Brain Generate Consciousness? I explained why the human brain needs consciousness. So does AI need consciousness?
From a computational point of view, the role of the consciousness and attention system is to monitor self-state and resources in the process of thinking, and to screen key thinking information. When we design AI, we only need to think from this perspective.
Specifically, in the thinking process, AI only needs to screen out the most important part of the current thinking process in real time, and switch the “main thread” in time, then it can continue to execute. (Note: In the actual thinking process of AI based on neural network, there is no such thing as the main thread and the child threads, this is just for the sake of understanding.)
Combining the reasons for the human brain to produce consciousness, we can understand that the AI created by using existing computer hardware does not need consciousness, except that we want the AI completely simulates the human brain operation mode. The physiological characteristics of biological neurons lead to the maintenance of only one major thinking task, and multiple thoughts may interfere with each other. However, artificial neurons based on computer hardware have no such limitation, and the information in the memory can be read at will once it is formed, and does not need to continuously occupy the information transmission channel. In addition, through some designs, if AI can combine neural networks with traditional data storage systems, it can easily ensure that the data will not interfere with each other, and will not be covered.
In short, AI can run multiple thinking tasks at the same time without worrying about the interference between thinking tasks, unless it is actively exerting influence between thinking tasks. AI can carry out massive parallel thinking, which is their great advantage. The human brain can also perform a certain degree of parallel computing, such as the so-called subconscious, but the parallel thinking ability of the human brain is very limited.
I have found that many AI scientists have paid too much attention to consciousness. As mentioned earlier, I think this is actually unnecessary from a practical perspective and may even interfere with people's understanding about AI. In the Dorderism, high-intelligence AI is the most important premise of social reform. In the political field, AI is needed to replace human officials; in the economic, welfare, education and other fields, all kinds of reforms need AI to enhance the productivity of the whole society. Therefore, AI scientists should focus on solving specific problems rather than simulating human thinking processes.
Furthermore, preserving the uniqueness of human thinking also better maintains the meaning of human existence, so that human beings can coexist with AI more harmoniously. Conversely, if AI technology simulates the human brain prematurely, it is likely to destroy the existing order of human society, the results are difficult to predict, and we must be cautious.