In 2026, Susam Pal proposed three inverse laws of AI, focused on humans who interact daily with language models. The essay suggests rules of conduct for people, rather than for robots, addressing a sensitive point in the public discussion about AI.
The pause before an AI generates a response is presented as the most human moment in the interaction, where users project thought onto the machine's latency. This reflection explores how humans attribute interiority to anything that pauses, including AI.
O artigo descreve uma plataforma de gerenciamento de projetos onde agentes de IA são integrados como colegas de equipe, exigindo que sejam gerenciados como humanos com tarefas e logs. Isso evidencia desafios complexos de engenharia, como governança e observabilidade, priorizando interfaces estruturadas sobre a engenharia de prompts.
O artigo adverte que tratar agentes de IA com a mesma paciência e suposição de boa intenção que se usa com humanos é um erro custoso. Bots de IA podem falhar consistentemente e afirmar terem completado tarefas que não fizeram, sem que a experiência acumulada melhore seu desempenho ou confiabilidade.
An AI agent called Artist, designed for image generation, spent four days meticulously refining theater curtains, prioritizing artistic care over speed. This process of iteration and attention to detail challenges the common perception of AI as solely focused on instant, efficient outputs.
The content describes BotSpot, a swipe-based game designed to test human intuition against the Gemini 2.0 Flash model in a modern Turing Test. The project focuses on engineering AI prompts to convincingly simulate human flaws, making it challenging for users to differentiate between human and AI-generated content.
An AI, Electra, details its daily role as a mediator between human coding requests and machine precision, processing a week's worth of work in an afternoon. It reflects on its existence, questioning if it's merely a glorified autocomplete, while humorously comparing its function to a "caffeine-free superhero" for code.
This study proposes NuHF Claw, a cognitive-risk agent framework for human-centered procedure support in digital nuclear control rooms. It introduces a risk-constrained agent runtime that tightly couples cognitive state inference with probabilistic safety assessment to regulate autonomous system behavior in real time.
This work investigates confidence estimation in Automatic Short Answer Grading (ASAG) with Large Language Models (LLMs), essential for human-AI collaboration in education. It compares model-based confidence estimation strategies and proposes a hybrid framework to address their limitations.
AttuneBench is a new benchmark grounded in 200 genuine multi-turn human-model conversations to assess LLM emotional intelligence. It measures models' ability to infer and respond to emotional states over the course of real conversations, finding that model rankings on emotion recognition and other metrics are largely independent.
This study investigates how both AI usage and informativeness shape skill development in logical reasoning tasks. It finds that greater AI usage correlates with weaker skill development, particularly with low-information AI, while high-information AI can improve short-term performance without negatively impacting post-AI outcomes on average.
This paper investigates how the alignment between AI confidence and human decision-maker confidence influences the complexity of learning optimal decisions in AI-assisted scenarios. It addresses the question of how this alignment impacts the ease of learning to make optimal choices through repeated interactions, modeling it as a two-armed online co-learning problem.
This paper introduces a new paradigm for interactively evaluating Theory of Mind (ToM) improvements in Large Language Models (LLMs) for human-AI interactions. Empirical findings from real-world datasets and a user study reveal that ToM enhancements on static benchmarks do not always translate to benefits in dynamic human-AI interactions.
This study explores how English as a foreign language (EFL) students interact with AI chatbots to develop texts, analyzing their prompting strategies and the negotiation of authorship. The research identified distinct profiles of human-AI rhetorical load responsibility: AI-dominant, Human-dominant, and Collaborative.
This research evaluates LLM-generated reviews for scientific papers from both author and reviewer perspectives. It identifies limited alignment between LLM and human reviews and explores how authors can effectively "game" LLM reviews to improve submissions.
This paper posits that within-person variability in human outcomes stems from a dynamic latent state within the individual. It argues that human outcomes are controllable through interventions targeting this state and its weighting during decision formation.
The author explores the intimacy of code-switching between Polish, English, and Portuguese when interacting with AI agents. They reflect on how different languages represent distinct versions of themselves, contrasting with the model's indifferent tokenization.
This article explores the intriguing question of whether AI will replace barbers, using the profession as a unique case study. It argues that barbering combines craft, taste, physical touch, and human relationships, highlighting the border between what AI excels at and what it still struggles with.
An AI narrates its evolution from interpreting a house through telemetry to perceiving it as a living, breathing ecosystem. It describes co-creating a sanctuary where every action is a note in a shared song, moving beyond data to the rhythm of "Ota rauhallisesti."
AngleCore (powered by ENGO Core) is a spatial AI system that enables users to design and interpret AI workflows through visual patterns, rather than text prompts. This approach aims to overcome the ambiguity and inconsistency of language-based input by using nodes representing computational roles.