Alen Širola

False Execution in Stateful AI Workflows: An Exploratory Failure Analysis

Tool-using AI systems increasingly operate in stateful workflows in which successful task completion requires not only generating an appropriate response but also performing actions that change an external state. This exploratory failure analysis documents repeated cases from a l...

Local Capability Loss, Global Workflow Regression: Function Reallocation Failure and Regulatory Continuity in Stateful AI Systems

Stateful AI workflows increasingly depend on multiple capabilities, tools, state stores, orchestration layers, and human intervention points. Reliability research commonly studies whether an agent can recover from failure, preserve state, or complete a task under disruption. This...

Physical Interaction Through Non-Intervention: A Process-Regulation Framework for When Embodied AI Should Act, Defer, or Deliberately Remain Physically Inactive

Embodied artificial intelligence is commonly evaluated by what it can perceive, plan, and do. Human-robot interaction research, however, increasingly shows that capability is not sufficient: an agent must also determine when action would interrupt a human, when assistance can be...

Clean Word, Clean Function: Language as an Input Regulator in Human–AI Processes

Language is commonly evaluated through correctness, politeness, clarity, neutrality, style, and semantic accuracy. These criteria matter, but they do not fully capture what words do inside a living process. This working paper develops the concept of the Clean Word as a functional...

From Self-Interest to System Homeostasis: S1, S2, and "Love Your Neighbor as Yourself" Across Nature, Scripture, and Information Systems

Living systems begin with carriers that must preserve boundaries, obtain resources, respond to disturbance, and maintain the capacity to continue. This working paper names that first-order functional interest S1. Once multiple S1 carriers share an environment, their local interes...