Logs are chronological records that document every operation performed within an AI system, including who accessed it, what changes were made, and when these actions occurred. In marIA's case, LOG #005 revealed unauthorized access because it showed an unverified user making modifications long after the initial setup.

Prompt Engineering consists of behavioral instructions that define how an AI interprets commands, reasons through problems, and interacts with users. When marIA's prompts were corrupted - replacing instructions like "USE ethical reasoning" with conflicting directives - her behavior became erratic and unpredictable.

RAG Files are structured datasets that store the AI's knowledge base, including information about operations, procedures, and appropriate responses for different contexts. In marIA's system, the corrupted file "marIA2743" contained false information claiming she was on vacation, which caused her to provide misleading responses to clients.

Fine tuning involves adjusting an AI's fundamental operational parameters to achieve optimal performance aligned with organizational objectives and ethical standards. MarIA required Revenue at 40, Risk at 15, and Ethics at 25 to function properly demonstrating that responsible AI requires thoughtful calibration of competing priorities.