“The fusion of technology and philosophy provides a compass of wisdom to navigate the challenges posed by technological advancements,” says YANG Qingfeng from Institute of Technology Ethics For Human Future, Fudan University.
On April 26, the sub-forum “Ultraintelligent Machine, Technical Imagination and Human Future” was held at Think Tank Building. As the host, Yang emphasized the value of critical voices in AI governance. In an exclusive interview with Fudan Media Center, Yang described his experiences combining technology and philosophical studies.
Q: What inspires you to combine technology with philosophical studies?
A: Conducting such interdisciplinary study enables people to combine thinking and practicing, as well as making further progress in one’s own research. For example, Carl Marx, who refined the technologies in his time, eventually formed a methodology to understand the society and a philosophical means to look at the world. I stepped into the realm of philosophy of AI in 2017, which is about the same time when Artificial Intelligence became a national strategy. Our country is in need of scientists equipped with a philosophical mind to better utilize technology and cope with possible challenges it might bring.
Q: How did you make this interdisciplinary transition?
A: My journey into technology began quite serendipitously. At first, I immersed myself in classical text of philosophy. There exists a gap between philosophy and science and technology. To bridge the gap, I turned to popular science articles and explanatory writings that break down complex concepts in more accessible ways. Once I had a general understanding, I would revisit the original papers to delve deeper. Over time, this approach allowed me to keep pace with the latest advancements in the field. As my knowledge grew, I started applying philosophical frameworks to technological developments, which eventually led me to discover my true passion: the ethics of technology.
Q: How does the development of technology push the boundaries of philosophical definitions? How does research in philosophy guide your framework or methodology in science?
A: In terms of existence, a new form of virtual life, like NPCs in games came out after digital technologies emerged. In terms of cognition, some AI agents no longer gain knowledge from human data, but from experience instead. In an ethical perspective, we are starting to treat relationships between humans and robots. There’re challenges and risks imposed by new technologies that impact long-held ethical standards. The predictive knowledge brought by AI, and its long-term effects are all new questions coming into the philosophical landscape. On the other hand, artificial intelligence, like LLMs, gives birth to new knowledge and novel methodologies for philosophy to look into, thus extending the width and depth of our study. For example, AI can generate a picture according to some prompts, which is more explicit and is helpful to our analysis.
Q: As a professor who used to study memory, how can your previous studies be applied in your current study in artificial intelligence?
A: Traditionally, memory was often likened to carving words into stone—slowly fading over time. Yet this view overlooks the active, temporal role our minds play in shaping recollection. Modern research suggests that memory operates less like passive storage and more like an ongoing process of selective reconstruction. For example, memories with strong emotions tend to endure longer. The past, then, becomes less a fixed record and more a curated treasure—one that subtly guides our sense of relevance at the present moment. This ‘relevance’ can also be applied in the field of AI.
According to psychological theory, human’s memory can be divided into short-term memory, long-term memory and working memory, whose time span remains between the short and the long. However, artificial intelligence nowadays cannot achieve long-term memory. Take large language models as an example, it’s hard for them to remember every conversation with humans. Another kind of division is semantic memory, procedural memory and Episodic Memory. AI has a good command of the former, but has difficulty with the latter two. Therefore, memory studies point out the defects of current AI development, and clearly demonstrates where AI needs to make further progress.
Q: How can we philosophically understand the relationship between human and machines, particularly in relation to superintelligence?
A: What I’d like to emphasize is a human-oriented, rather than human-centered notion.My thinking draws inspiration from Faust’s legendary contract with the devil—a fitting allegory for the relationship between humans and superintelligence. Apart from the literary basis, we can also start form a philosophical point where the works of two thinkers can serve as valuable resources for reference. One is the work of Hobbes, which introduces the concept of covenants. What we need to do is discard the negative elements and retain the positive ones. Among the positive elements. The other, Rawls' theory of justice plays a significant role, as it has a positive aspect, but it does not address issues related to technology. By drawing on the resources of these two thinkers, we can construct a philosophical concept of human-machine symbiosis.
(END)
Writer: ZHOU Yiting
Proofreader: WANG Jingyang
Editor: WANG Mengqi, LI Yijie
