Ge Jianxiong: In the AI era, humanity cannot hand value judgments over to AI

"I used to see depictions like this in science fiction, and at the time I thought, how happy such a day would be. But now I'm a bit worried." Recently, at a lecture themed "Science, Humanities, Life—How to Face the AI Era" held at the NYU Shanghai Library, Ge Jianxiong, a renowned historian, Distinguished Senior Professor of Liberal Arts at Fudan University, and Director of the Library of The Chinese University of Hong Kong, Shenzhen, mentioned a scene he had once read in a book: waking every morning to soft music, curtains opening automatically, breakfast already on the table; arriving at the office, the day's work list already prepared, and with a few button presses, the day's work is done. "If that day really comes, may I ask—what pleasure would be left in our lives?" He believes that humanity's value judgments, thoughts and feelings, and beliefs and convictions can never be handed over to AI. "What guides humanity's future is not any science or technology, nor AI, but our values, our beliefs—it is the humanities." The "artificial" in AI is real, but the "intelligence" may not be Ge Jianxiong began his discussion of AI with the two main threads of human history. "One is material progress, the other is progress in human nature and the humanities." He said material progress always means today surpasses yesterday and tomorrow surpasses today, accumulating continuously; but the humanities do not follow such a law—the peaks of thought reached by those geniuses in human history may never be surpassed. On this basis, Ge Jianxiong divides knowledge systems into two major categories, science and the humanities: science is based on the material, has standard answers, and requires empirical evidence and reproducibility; the humanities are based on the spirit, have no objective standards, and do not require verification. "Precisely because of this, what should lead the future is the humanities, not some specific science or technology." So where does today's AI stand? "The AI (artificial intelligence) we talk about now—the artificial is real, but that intelligence is not real; it is merely a simulation." Ge Jianxiong gave an example of an experiment a student did: have AI write two papers, one on unity and division in Chinese history, which it wrote very well—because there is so much material and data on this; the other on the population of the Western Han, which it wrote very poorly—because papers and books worldwide on the Western Han population are few and far between, and to fill space, AI fabricated a lot of data. "AI can integrate all existing information, but the premise is that this information has already been digitized." Ge Jianxiong said that the never-publicly-disclosed unique manuscripts in libraries, content protected by intellectual property, state secrets, and personal privacy are all beyond AI's knowledge. More importantly, facts can be digitized, but evaluations cannot. "AI can clarify all the facts of Qin Shi Huang's life—the day he was born, the day he died, what he did—more comprehensively than any historian. But as for how to evaluate Qin Shi Huang, are we to rely on AI?" Ge Jianxiong then asked, "For another example, humanity has different beliefs, and beliefs do not require reasoning. After using AI, are we to obey AI's judgments? Must the consolidation of belief also rely on AI?" He mentioned that perhaps in the future, after people are born they can live entirely relying on robots and AI technology, encounter no difficulties, and not even need to think for themselves, living easily to 200 or even 300 years old. "But how would that be any different from other living things? Where is the value of being alive?" He believes people need pursuits beyond the material, to maintain their own aesthetics and sensibilities, and even to actively undergo difficulties. Like those who train for triathlons or athletes pursuing better Olympic results, they may not realize any material value, but they pursue a spiritual value, a value of life. Don't let the brain "degenerate" like the appendix As a scholar who has long dealt with young people, Ge Jianxiong is especially vigilant about AI entering campuses. "People's memory, critical thinking, logical reasoning, and language abilities all depend on continuous self-training. If one relies on AI too early, these intelligences will remain in infancy, or even degenerate." He made an analogy: the human appendix and gallbladder have degenerated because humans no longer eat raw food and food has become increasingly refined. "Physiological degeneration is still a small matter. If in the future the functions of the human brain also degenerate like this—if one day the AI system has a problem, stops, or is destroyed, how will these people survive?" In the dialogue, NYU Shanghai Chancellor Tong Shijun supplemented this view from another angle. He cited a pair of concepts from philosophy: "having knowledge" and "knowing." "In terms of having knowledge, AI can indeed play a huge role; but in helping people personally experience knowing and understanding, AI's role is probably destined to be very limited." Tong Shijun admitted frankly that the dilemma of today's education lies in the fact that social evaluation compares precisely the speed and quantity of "having knowledge," forcing students to use AI. "So merely reminding everyone to 'use less AI' is not enough; educational evaluation and the social system itself must both be adjusted." NYU Shanghai Chancellor Tong Shijun (right) in dialogue with Ge Jianxiong. Ge Jianxiong responded: in which direction AI develops, and which uses are appropriate and which are not, this judgment cannot be decided by technology itself; it must be guided by humanistic values. "Just as gene editing has ethical prohibitions, the development of AI must also achieve balance through comprehensive social consideration." At the event, a first-year student at NYU Shanghai expressed confusion: "If the future is precisely predicted by AI, what is the meaning of life?" Ge Jianxiong said that the more advanced technology becomes, the more people must maintain subjectivity and conscience. "Even if AI is even more powerful in material and data prediction, in spiritual and value choices, we can still choose to believe in ourselves. Pursuing spiritual value and challenging oneself is itself the meaning of life." At the end of the lecture, Tong Shijun quoted a passage from Mencius, King Hui of Liang I: "To tuck Mount Tai under one's arm and leap over the North Sea, and to say to others, 'I cannot'—this is truly being unable. To break a branch for an elder, and to say to others, 'I cannot'—this is not doing it, not being unable." Mencius used this to distinguish the question of whether one is able from the question of whether one acts. Tong Shijun then said that even if in the future AI is capable of everything, humanity must still think clearly—which things in principle should not be done, and should not be handed over to any external thing.