When will general artificial intelligence arrive, where will it take mankind, and how will humans and AI coexist? Although there is no answer at this stage, more and more people have put forward their own ideas. Recently,Well-known science fiction writer Han Song raised a question in the Tencent News "20 People in 20 Years 20 Questions" program - "What impact will artificial intelligence and the metaverse have on gender relations in the future? Will the concept of family disappear?"
The essence of Han Song's question is to ask - when AI has autonomous consciousness and emotional capabilities, what kind of social relationships may it form with humans?
Regarding the issue of AI consciousness, there are different debates in the industry, and large companies are very cautious and sensitive about this.
In 2022, Google fired a senior engineer due to the controversy caused by the awakening of AI consciousness. The engineer's name was Lemoine. When he was chatting with the chat robot developed by Google based on LaMDA, he discovered that,The AI has the intelligence of an 8-year-old child and has a soul. After feeding back relevant information to the company, it was asked to take a paid leave. Lemoine made the relevant information public and was eventually suspended by Google for violating the confidentiality agreement.
Lemoine is just a radical epitome of the group that supports the awakening of AI consciousness. Tammy Besiroglu, a visiting research scholar at MIT, was once counterattacked for drawing the line on consciousness fed back from existing large models.
Based on the model training data cited by Besiroglu, Besiroglu believes that the models above the black implementation in the table have weak consciousness.
Also in the "20 People and 20 Questions in 20 Years" planning, in response to Academician Wu Hequan's question about what level AI technology will reach in the next 20 years, Yuan Yang, assistant professor at the Institute of Cross-Information at Tsinghua University, also talked about the topic that AI will have self-awareness.
In addition to supporting the awakening of AI’s consciousness, Yuan Yang even believes that in the next 20 years, AI’s consciousness can be separated from its “body”.
In Yuan Yang's view, the success of the large model has verified the hypothesis that intelligence and body can be perfectly separated in the future, and AI intelligence is more stable, easier to copy and adjust than human intelligence. "Animals on the earth are all individuals who are responsible for their own profits and losses. The energy of the brain needs to be supplemented by the body's foraging. This will give us the illusion that a living body must have a physical body before it can have self-awareness."
What is AI’s self-awareness? Yuan Yang believes that it depends on what signals the AI receives and what tasks it needs to process. In the future, AI can even have the self-awareness of an organization.He emphasized that if people connect an ontology to AI and equip it with many sensors, then AI will easily regard the ontology as part of itself. In the future, AI can even map the ontology to a larger organization, such as a building.
Consciousness is one of the important prerequisites for the generation of emotions. When AI has true autonomous consciousness, it is the basis for us to talk about the emotions between humans and AI, and even to formulate topics. Although humans have already made many assumptions in this regard, and even some related products are gradually being implemented, even though they do not yet have true consciousness.
Musk dancing with the robot "female partner" output by AIGC, source: Internet
In 2022, Musk had a conversation about the "future" with TED host Chris Anderson. When asked whether "robots might become romantic or even sexual partners," he replied, "It's probably inevitable. I mean, I did promise online that I would make a 'Catwoman,' and we can make a 'Catwoman' robot."
Musk's "Catwoman" is another vision of the movie "Her" in reality. Going one step further, some developers are already trying to create apps related to emotional companions.
In 2013, a science fiction romance film "Her" written and directed by Spike Jonze was released. The film tells the story of the writer Theodore who falls in love with "Samantha" in the computer operating system. Her voice is sexy and her expressions are humorous and considerate.
Going one step further, some developers are trying to create a virtual human clone based on a real person's image, transforming into an AI girlfriend chatbot.
In response to writer Han Song's question, Chen Qiufan and Chen Chen, both science fiction writers, gave their own thoughts. In their view, traditional family concepts will face a strong impact in the future.
Chen Qiufan believes that artificial intelligence and the metaverse can provide an unprecedented interactive platform that transcends the limitations of time and space. "Gender relationships may no longer be limited to physical existence, and emotional connection and spiritual resonance may become dominant."
In Chen Qiufan's view, in science fiction novels we often see the development of deep emotions between humans and robots, humans and virtual characters that transcend the limitations of reality, such as Philip K. Dick's "Do Androids Dream of Electric Sheep?" The complex emotional entanglement between humans and androids in "The Matrix" or the blurring of the boundaries between virtuality and reality in "The Matrix" are highly consistent with Professor Yuan Yang of Tsinghua University's view on the "separation of body and consciousness" of AI.
As for the concept of family, Chen Qiufan believes that the traditional family model will be overturned. Individuals will have multiple choices to freely connect into new social functional structures. In the future, families will present diverse relationship forms, such as more and more "fluid partnerships" or "flexible partnerships".
Science fiction writer Chen Qiufan agrees with Chen Qiufan on the subversion of the family concept. She also believes that "the impact of the virtual world on the family will be relatively stronger, because it will give humans a second or even third level of electronic living space at different levels and positions. This will bring uncertainty to individuals' identities, and also means uncertainty in emotional and ethical relationships. Fixed partners or family relationships will disappear as a result, and will be replaced by relationships that are temporary or have complex agreements."
However, Chen Chen also gave some calm thoughts. Although it has super subversive capabilities, these are all based on the premise of high automation. However, we have no timetable for human beings to enter this era of automation.
“Artificial intelligence AI is still an automated system that serves humans.As the system's learning capabilities continue to improve and the degree of automation becomes higher and higher, it will eventually get rid of the program constraints set by humans and become strong artificial intelligence, which is the kind of self-aware artificial intelligence that often appears in science fiction novels.By that time, robots with artificial intelligence will become completely independent thinking individuals, may have human-like emotions, and intervene in human gender relationships and family life. "
Humans have made many assumptions about AI, but there are also paradoxes - when AI has autonomous consciousness, there is no standard answer to this question: whether AI will develop family and sexual relationships with humans, and whether AI needs emotions.
"The robot's appeal should be based on its own intellectual development needs. It does not require "human emotions"," Chen Chen added.
Regarding the research on this issue, President Huang Tiejun of the Intelligent Source Research Institute has repeatedly emphasized that if AI intelligence surpasses humans in an all-round way, that is, if AGI is achieved, humans will not be able to control it.Anthropocentrism only applies to AI whose intelligence is lower than that of humans. It is the "geocentric theory" and "heliocentric theory" in the field of intelligence.To make AGI consistent with human goals and values, the initiative is not on the side of humans.
In Huang Tiejun’s view, when human intelligence is higher than AI intelligence, AI is a controllable assistant that can be trained by humans to become an increasingly trustworthy assistant. When AI intelligence is higher than human intelligence, that is, when AGI (artificial general intelligence) that comprehensively surpasses humans appears, anthropocentrism is shattered, and the question becomes whether AGI believes in humans, not whether humans believe in AGI.
In other words, when AI becomes autonomous and has the right to make active choices, whether it wants to start a family with humans or whether it should trust humans, perhaps it should also listen to AI’s ideas instead of humans’ “wishful thinking.”