Yesterday, before leaving work, I discussed the impact of chatGPT with my colleagues, which led to this "rambling".
This topic has been widely discussed by people from all walks of life recently. As someone who has been involved in the field of artificial intelligence for over ten years, I was also shocked by chatGPT.
The pace of technological advancement far exceeds that of humans. Half a year or a year ago, we were generally concerned that the AI boom was coming to an end. Most technologies in the fields of image, video, translation, and speech had already been implemented. The technologies that had not been implemented were still in the research and development stage, and it might take another 5-6 years or even more than 10 years for them to become available, following the historical pattern of previous winters. However, the emergence of chatGPT seems to have changed this perception. The appearance of chatGPT and midjourney has injected a shot of adrenaline into the development of artificial intelligence, and this shot of adrenaline has not only been given to industry insiders but also to almost everyone in society. I once saw an article discussing chatGPT on the WeChat public account "Exploration and Debate". Overall, because it involves the field of human language, which is considered a symbol of wisdom, people both love and worry about chatGPT. They love it because it can improve productivity, but they are afraid that it will take away job opportunities.
When I chose my college major more than ten years ago, I had never heard of the term "artificial intelligence". I wanted to study intelligent science and technology. In the end, I accidentally studied automation, which happened to have a direction in artificial intelligence. At that time, artificial intelligence was not a popular discipline. There was very little data, limited computing power, and no good applications. Go was considered an impossible task, and research in automation was all about "pattern recognition". When I was choosing my specialization in my third year, out of hundreds of students in the department, less than 10 chose the direction of artificial intelligence.
In the first few years of my graduate studies, artificial intelligence gradually became known to more people due to deep learning. By the time AlphaGo came out in 2016, artificial intelligence had become the most popular major. The neural network course taught by my former advisor went from having only 30 students to not being able to accommodate all the students in the largest lecture hall on campus.
From my insignificant experience, I can clearly feel the arrival of the artificial intelligence boom and the rapid progress in this field. It seems that in an instant, previously unattainable fields such as image recognition, machine translation, and speech recognition have become ordinary technologies that are everywhere.
Therefore, after the appearance of chatGPT, even though it currently has various shortcomings, we worry about what the situation will be like in 1 year or 5 years. We can't even imagine what it will be like decades from now.
We discussed the issue of consciousness. It is generally believed that machines do not have consciousness, they do not have their own thoughts, and they only provide answers according to our needs. But what if we give them a very broad question? For example, if a robot integrates chatGPT and we tell it, "You need to build an office building. You need to independently purchase the materials required for construction through programmatic structures, generate control programs to control body movements for physical labor, and fully understand the intentions of humans in order to persuade them." In this situation, chatGPT can completely behave like a conscious being and gradually achieve our goals according to its own designed plan. From "omniscience" to "omnipotence," we don't know how far it can go.
What is creativity? Some people say that creativity is the combination of various ideas. According to this definition, chatGPT is a creative master, generating its own interpretations based on the knowledge of the entire Internet. Similar to creativity, humans have "flashes of inspiration" and "emergence". ChatGPT may not have these sudden inspirations for the time being, but we don't know if they will appear in the future.
Researchers may be safe for now because chatGPT is trained on existing human knowledge, and the goal of researchers is to discover knowledge that humans do not yet know. Therefore, the research field is still safe until chatGPT can independently conduct scientific research and discover new knowledge. But how long this safety can be maintained, I think everyone is uncertain.
Although many people oppose the further development of technologies like chatGPT, just like Bitcoin, once a technology appears, it will not disappear for no reason. We should still think more about how to use chatGPT rather than how to block it because we cannot stop it.