Tag
The article discusses the ethical boundaries of AI agent autonomy, questioning which actions should require human approval even if technically feasible, using refund decisions as an example.
The article argues that humans must stay actively engaged when using LLMs, as automation has limits in complex tasks requiring critical thinking and oversight.
The article explores whether AI accelerates individual tasks while shifting bottlenecks in broader workflows, urging measurement of end-to-end process gains and considering improvements in tool connectivity, ownership, and steps.
The author discusses testing an AI system for product identification and listing in reselling, exploring how to determine when AI should act independently or escalate to humans, emphasizing a process of investigation and verification.
Introduces READY, an evaluation framework for qualifying AI agents for enterprise deployment by measuring reliability, human oversight burden, and cost, enabling evidence-based deployment decisions.
This article describes how Sabrina Ramonov set up a team of four AI bots using Grok Bot for marketing tasks, highlighting rules for human oversight and bot limits.
This paper presents AIMC, a visual analytics framework for human oversight of autonomous scientific discovery, enabling monitoring and understanding of AI-generated research artifacts.
The author argues that understanding how AI works is more valuable than just learning to use it, illustrated by a clinic's failed AI implementation that caused claim denials, and emphasizes the need for human oversight in AI systems.
The paper argues that current AI agent systems degrade human oversight capabilities and advocates for design-level changes to support effective human-agent interaction, emphasizing the need to maintain human skills.
The article explores how AI's greatest risk isn't rebellion but human complacency, where people stop verifying AI outputs due to perceived reliability, leading to unchecked failures.
A discussion of how ready AI agents are for real-world work, covering their current abilities and the key open questions around reliability, permissions, failures, and human oversight.
This article examines the gap between hype and reality for AI coding agents, arguing that they are effective for accelerating workflow parts but still require human oversight for architecture, debugging, and review, and questioning whether current benchmarks measure the right things.
Ford rehired hundreds of veteran engineers after its aggressive AI adoption led to costly quality issues. The automaker now uses AI alongside human oversight to improve production quality.
A commentary emphasizing that despite AI advances, human understanding remains crucial for safe and humane deployment, urging users to verify AI outputs and treat AI with respect.
This paper proposes that reliability in AI-assisted social science research depends on decision architecture—how cognitive labor is divided between humans and machines. Through a pre-specified factorial experiment, the authors show that an unconstrained multi-agent baseline fails in 72% of runs, while one organized with three architectural commitments (LLMs restricted to reasoning, deterministic data/estimation, and three human decision gates) fails in only 16%.
This paper analyzes 35,361 GitHub code comments referencing AI use to develop a taxonomy of AI-assisted development activities, finding that developers primarily use LLMs for code implementation and enhancement, with subsequent human refactoring and bug fixes, and a temporal shift toward conceptual support over direct code generation.
A discussion on the ethical implications of fully autonomous AI agents, focusing on accountability, decision-making, privacy, and human oversight.
This paper surveys the capabilities and limitations of AI across the full research lifecycle, from idea generation to dissemination, identifying a sharp boundary between reliable assistance and unreliable autonomy. It provides a taxonomy, benchmark suite, tool inventory, and design principles for human-governed AI collaboration in research.
The article discusses how companies can integrate EU AI Act compliance into their product development from the design phase, highlighting transparency, guardrails, and human oversight as key architectural changes.
This paper introduces the Functional Intentionality Test (FIT) and FIT-Eval framework to quantify the degree of intentional-like behavior in agentic AI systems for governance and accountability purposes.