This study aims to analyze the epistemological and ontological transformations occurring in the natural sciences in the era of artificial intelligence (AI). The method used is a qualitative approach with a case study design, using data sources from scientific literature and recent articles. Data analysis was conducted thematically to identify key patterns related to changes in the way scientific knowledge is acquired, validated, and understood. The results show that the emergence of AI has shifted the epistemological paradigm of the natural sciences, where the process of knowledge production and validation no longer relies entirely on human observation and reasoning, but also on the ability of algorithms to process large-scale data. This condition presents new challenges in the form of the risk of algorithmic bias, reduced scientific transparency, and ethical dilemmas related to the responsibility and accountability of research results. Ontologically, AI also creates new forms of reality through digital models and simulations that are often considered to represent scientific reality. This change demands an update in how the natural sciences understand the concepts of "reality" and "scientific truth." Therefore, an adaptive epistemological and ethical framework for technology is needed so that AI can be used responsibly without eliminating the role of humans as the primary subjects of knowledge. This research emphasizes that collaboration between humans and AI must be based on the principles of transparency, interpretability, and humanity to ensure that science remains credible, inclusive, and meaningful.