Unpredictability of AI
The young field of AI Safety is still in the process of identifying its challenges and limitations. In this paper, we formally describe one such impossibility result, namely Unpredictability of AI. We prove that it is impossible to precisely and consistently predict what specific actions a smarter-than-human intelligent system will take to achieve its objectives, even if we know terminal goals of the system. In conclusion, impact of Unpredictability on AI Safety is discussed.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Environmental unpredictability and inbreeding depression select for mixed dispersal syndromes
Mixed dispersal syndromes have historically been regarded as bet-hedging mechanisms that enhance survival in unpredictable environments, ensuring that some propagules stay in the maternal environment while others can pot…
On the Confidence in Bit-Alias Measurement of Physical Unclonable Functions
Physical Unclonable Functions (PUFs) are modern solutions for cheap and secure key storage. The security level strongly depends on a PUF's unpredictability, which is impaired if certain bits of the PUF response tend towa…
Diversity Boosts AI-Generated Text Detection
Detecting AI-generated text is an increasing necessity to combat misuse of LLMs in education, business compliance, journalism, and social media, where synthetic fluency can mask misinformation or deception. While prior d…
Text DetectionComputational Phenomenology of Temporal Experience in Autism: Quantifying the Emotional and Narrative Characteristics of Lived Unpredictability
Disturbances in temporality, such as desynchronization with the social environment and its unpredictability, are considered core features of autism with a deep impact on relationships. However, limitations regarding rese…
Numerical Instability and Chaos: Quantifying the Unpredictability of Large Language Models
As Large Language Models (LLMs) are increasingly integrated into agentic workflows, their unpredictability stemming from numerical instability has emerged as a critical reliability issue. While recent studies have demons…