Sheng-Qi MATo most people, artificial intelligence seems to be a being that can replace humans. Even in many movies and novels, artificial intelligence is exaggerated to be omnipotent. Artificial intelligence has become synonymous with technology, speed, power and the future. I can't help but think, can artificial intelligence really replace human beings? And what is the difference between artificial intelligence and us?
To answer this question, we first need to think about what the biggest difference between us humans and artificial intelligence is. After reviewing a large number of papers by AI scientists and referring to Kaifu Lee's TED talk, I believe that the biggest difference between humans and AI is by far the one we take for granted —— humanity.
Let's first review what the definition of humanity is and I will use practical examples to explain these abstract concepts.
Definition of humanity: compassionate, sympathetic, or generous behavior or disposition
Facebook's AI often suggests posts to users, but it doesn't recognize which posts are good and which posts are bad. As we know, the posts of Facebook which about violence, hatred, discrimination often have many coments, artificial intelligence will mistakenly think that this is content that humans like, so artificial intelligence pushes these posts to more people. A Facebook researcher registered a new Facebook account in 2020 to try to understand the experience of ordinary Indian netizens using Facebook. But in the next three weeks, the researcher saw Facebook push a lot of hate speech, misinformation, and violent content. As of today, artificial intelligence can only identify 2 to 5 percent of hatred speech.
However, if the database built up by artificial intelligence is large enough that it should be no problem to conduct conversations and respond to retreats like human beings. The basic ethical and moral behaviors such as politeness, giving up one's seat, and obeying the law can probably be done properly depending on the environment and society in which one lives. But in some controversial situations, artificial intelligence will not be able to do so.
For example, in the famous tram problem, suppose there are 5 people tied to a tram track and 1 person tied to its spare track, and there is a runaway tram coming fast, and you have a joystick nearby, you can push the joystick to make the tram go into the spare track, killing the 1 person and saving the 5 people. You can also do nothing, kill 5 people and save 1 person. Obviously, we can't expect AI to do anything when humans themselves are unable to make moral judgments.
With the above two examples, we can believe that AI will only be a smarter machine, not a dumber human, for decades to come. The reason for this is that in addition to their inability to make moral judgments and their lack of humanity, the last and most important point I want to talk about is the love.
Regarding AI and love, I particularly like an example mentioned by Professor Kai-Fu Lee in his TED talk. Artificial intelligence is well known to us from Alpha Go defeated the world Go champion Ke Jie. Because humans once believed that Go represents the highest level of human intelligence and the machine can never reach that. When AlphaGo defeated the world champion Ke Jie, while Ke Jie was crying and loving the game of go, AlphaGo felt no happiness from winning and certainly no desire to hug a loved one.[1]
AI makes our lives smart and efficient, freeing us from routine jobs[2]. The future will see more AI that resembles humans, but instead of spending time discussing how not to be replaced, We should think more deeply about how to value the only gift we have as human beings—— humanity and love.
[1] : Kai-Fu Lee, 2018, How Artificial intelligence can save our humanity?
[2]: Kai-Fu Lee, 2018, TED Talk