Research & Publications
Research and publications by Luiza Scurtu of Life Laboratory, including work on artificial intelligence, human judgment, AI governance, human interaction and relational architecture.
Research publications
Before Judgment. How artificial intelligence can change the problem before a human
Author: Luiza Scurtu
Published: 30 August 2026 · Publication/resource type: Other
Human oversight is commonly placed near the end of an artificial-intelligence-assisted decision: a system produces an analysis and a person reviews it before action. This assumes that the human reaches judgment with an independently formed account. Yet artificial intelligence can participate earlier by selecting information, organising relevance, proposing causes and shaping the language through which a problem becomes intelligible. Final authority may remain human while the decision object has been partly constructed elsewhere.
This article proposes the concept of judgment-object divergence. It develops one specific mechanism, latent-intent hypothesis divergence, in which humans and artificial-intelligence systems generate, retain and weight different explanations involving concealed goals, incentives, status, deception, fear, loyalty, retaliation or strategic conduct. Recent research already establishes the broader territory of epistemic mediation, changed problem formulation, representational gaps and weakened epistemic agency. The contribution here is to connect those findings to the behavioural hypotheses that shape ambiguous strategic situations and to show why final human approval may preserve authority in form while failing to preserve independent judgment in function.
DOI: 10.5281/zenodo.22177289
Read or download the supplied PDF
The Three-Gate Architecture: Inspection, deployment and allocation in the governance of frontier artificial intelligence
Author: Luiza Scurtu
Published: 9 August 2026 · Publication/resource type: Working paper
Frontier artificial intelligence is acquiring governance arrangements before comprehensive legislation has assigned stable decision rights over its most consequential capabilities. This working paper examines whether recent developments are better understood by separating three distinct objects of control: authoritative knowledge about capability, the capability state that leaves controlled development, and entitlement to receive or retain that state. It terms the corresponding decision sites inspection, deployment and allocation.
The study combines a structured qualitative theory audit with a time-bounded, theory-building analysis of publicly documented frontier-AI cybersecurity cases in the United States and related allied responses through 9 August 2026. It argues that the contribution of the Three-Gate Architecture is not a new general theory of AI governance, but an integrative institutional topology that prevents distinct decision objects from being collapsed.
The paper also proposes a candidate mechanism — cross-gate translation under reciprocal dependence — through which an inspection judgment may become a deployment condition and differentiated deployment may generate recipient categories. The evidence strongly supports the presence of all three gates, provides moderate support for cross-gate coupling, and does not establish regime formation.
The framework is intended as a diagnostic and comparative map of decision rights in frontier-AI governance. The mechanism remains provisional pending repeated, traceable cross-gate causation across model cycles, capability domains and jurisdictions.
DOI: 10.5281/zenodo.21860649
Read or download the supplied PDF
Baseline Drift: AI-shaped conversational standards and human interaction
Author: Luiza Scurtu
Published: 20 July 2026 · Publication/resource type: Report
This independent conceptual analysis and research agenda examines whether repeated interaction with rapid, patient, and adaptive artificial-intelligence systems can recalibrate the standards used to experience ordinary human conversation.
It assesses the proposed pattern, labelled Baseline Drift, against adaptation-level and range-frequency theory, sequential contrast, expectation–disconfirmation, processing fluency, conversational timing and repair, behavioural spillover, norm leakage, and the machine-integrated relational adaptation model.
The updated evidence provides causal support for the broader proposition that human–AI interaction can spill over into later human-directed judgment and behaviour across several experimental contexts. However, no study identified in this targeted review isolates a durable, feature-specific recalibration of tolerance for human delay, hesitation, ambiguity, or repair. Baseline Drift is therefore retained as a provisional descriptive hypothesis rather than a distinct psychological mechanism.
DOI: 10.5281/zenodo.21458564
Read or download the supplied PDF
Attachment by Design: AI attachment under provider-controlled relational architecture
Author: Luiza Scurtu
Published: 20 July 2026 · Publication/resource type: Technical note
This independent theory audit and mechanism analysis examines how attachment to artificial-intelligence systems is explained by established research on attachment, object attachment, parasociality, social presence, anthropomorphism, perceived responsiveness, need satisfaction, habit, trust, commitment, and platform power.
The analysis finds that existing theories explain most attachment formation. “Attachment by Design” is therefore presented as a provisional integrative framework that distinguishes four layers: attachment formation, intensification through system affordances, reliance or dependence, and provider governance. No distinct elementary psychological mechanism is established. The report’s primary contribution is to integrate these processes within a provider-controlled relational architecture and identify the resulting research and governance questions.
DOI: 10.5281/zenodo.21454745
Read or download the supplied PDF
Book
The Fog Between Decisions: Why Leadership Feels Harder Than It Should – And What to Do About It
Luiza Scurtu · 2026 · Independently published
View the book