Why do you care about AI Existential Safety?
AI Existential Safety is extremly important because as AI systems become more advanced, they may be able to perform tasks that have significant consequences in the real world. If such systems are not designed and tested carefully, they could potentially cause harm or make mistakes that have serious consequences. By focusing on AI existential safety, we can help ensure that AI systems are beneficial to society and do not cause unintended harm. I have two clues, one practical, and the other philosophical.
1. My investigations on large language models had shown the possible dangerous feedbacks from these models without possible interpretations and the potential adoption of these models in real world systems could bring out the harmful results for humanity. To keep humanity sustainable, we should pay more attention to AI existential safety.
2. AI existential safety research also has many deep influence on our consideration of the future of humanity, the shape, development and the real meaning of existence, which are very interesting topics needing exploration through the intersecting perspectives of people with different backgrounds and culture.

Please give at least one example of your research interests related to AI existential safety:
I am working on ontological anti-crisis theory combining model theory, game theory and computational
complexity theory to give theoretical foundation for ontological crisis problem.
OAC theory put ontological crisis problem through a more theoretical view. I design the bounds and dynamics of the development of ontology for a model and made completeness and incompleteness results based on the construction which can be an analysis foundation for large models.
Some related constructions and design for OAC theory.

1. onto Bounds
   We consider bounds of ontology for specific task or problem:
   1. upper bound of ontology for a task
   2. lower bound of ontology for a task
2. onto Dynamics
   1. parallel ontology
   2. ontology intersecting
   3. one ontology absorbs another ontology
   4. ontology evolution
   5. meta ontology
3. onto Actions
   build / learn / reset / collapse

With solid mathematical foundations and deep results from TCS/Logic, more building blocks could emerge to help us dealing with ontological crisis problems and many other related AI existential safety problems, which is also a bridge for many researchers and students from different areas to work together for a better future with safe AI.