How do we determine whether an AI system has preserved a researcher's conceptual framework across repeated interactions?
My new article introduces the research behind *Prior-Data Gravity and Framework Reabsorption*, a preprint examining how large language models retain familiar terminology while changing the conceptual relationships that define the original framework.
The paper develops a behavioral theory, proposes framework fidelity as a distinct evaluation target, and presents a coding grid for examining structural preservation across corrections and subsequent generation.
#ArtificialIntelligence #GenerativeAI #LLM #AIResearch #MachineLearning #naturallanguageprocessing https://www.linkedin.com/pulse/when-ai-preserves-words-rewrites-research-cristina-gherghel-cpf7f
My new article introduces the research behind *Prior-Data Gravity and Framework Reabsorption*, a preprint examining how large language models retain familiar terminology while changing the conceptual relationships that define the original framework.
The paper develops a behavioral theory, proposes framework fidelity as a distinct evaluation target, and presents a coding grid for examining structural preservation across corrections and subsequent generation.
#ArtificialIntelligence #GenerativeAI #LLM #AIResearch #MachineLearning #naturallanguageprocessing https://www.linkedin.com/pulse/when-ai-preserves-words-rewrites-research-cristina-gherghel-cpf7f
How do we determine whether an AI system has preserved a researcher's conceptual framework across repeated interactions?
My new article introduces the research behind *Prior-Data Gravity and Framework Reabsorption*, a preprint examining how large language models retain familiar terminology while changing the conceptual relationships that define the original framework.
The paper develops a behavioral theory, proposes framework fidelity as a distinct evaluation target, and presents a coding grid for examining structural preservation across corrections and subsequent generation.
#ArtificialIntelligence #GenerativeAI #LLM #AIResearch #MachineLearning #naturallanguageprocessing https://www.linkedin.com/pulse/when-ai-preserves-words-rewrites-research-cristina-gherghel-cpf7f