BenchGuard is the first LLM-based automated auditing framework for execution-based agent benchmarks, identifying 12 confirmed issues in ScienceAgentBench and matching 83.3% of expert findings on BIXBench at low cost.
Jansson and Steven M
8 Pith papers cite this work, alongside 938 external citations. Polarity classification is still indexing.
8
Pith papers citing it
938
external citations · Crossref
citation-role summary
background 3
citation-polarity summary
years
2026 8roles
background 3polarities
background 3representative citing papers
Metacognitive self- and co-regulation loops improve LLM agent performance in engineering design by mitigating fixation and enabling better exploration of design options.
On 2,520 programming tasks, matched Qwen general and coder models reliably raise Bloom cognitive demand but fail to lower it, so execution skill does not imply educational control.
AI improves brainstorming quality for general-purpose impact assessment but not specialized applications when it offers hints early and structures ideas later, based on workshop evaluations with 54 participants.
Brief2Design supports a multi-phased workflow for AI-assisted graphic design from briefs, increasing prompt diversity and requirement handling ratings but requiring more generation time than conversational baselines.
Purrsuasion is a negotiation game that surfaces satisficing and intent-attribution difficulties when students practice ethical data disclosure under real constraints.
NexusAI decomposes LLM inspirations into navigable functional fragments and abstractions to improve creative design space exploration, with a user study showing reduced cognitive overhead.
A within-participants study with design students found that sketch inputs to an AI ideation tool increased fluency but students still preferred text prompts, pointing to design choices that could better preserve reflective practice.
citing papers explorer
-
BenchGuard: Who Guards the Benchmarks? Automated Auditing of LLM Agent Benchmarks
BenchGuard is the first LLM-based automated auditing framework for execution-based agent benchmarks, identifying 12 confirmed issues in ScienceAgentBench and matching 83.3% of expert findings on BIXBench at low cost.
-
Supervising Ralph Wiggum: Exploring a Metacognitive Co-Regulation Agentic AI Loop for Engineering Design
Metacognitive self- and co-regulation loops improve LLM agent performance in engineering design by mitigating fixation and enabling better exploration of design options.
-
From Execution to Education: A Bloom-Aligned Framework for Measuring Educational Control in LLMs
On 2,520 programming tasks, matched Qwen general and coder models reliably raise Bloom cognitive demand but fail to lower it, so execution skill does not imply educational control.
-
When and How AI Should Assist Brainstorming for AI Impact Assessment
AI improves brainstorming quality for general-purpose impact assessment but not specialized applications when it offers hints early and structures ideas later, based on workshop evaluations with 54 participants.
-
Brief2Design: A Multi-phased, Compositional Approach to Prompt-based Graphic Design
Brief2Design supports a multi-phased workflow for AI-assisted graphic design from briefs, increasing prompt diversity and requirement handling ratings but requiring more generation time than conversational baselines.
-
Investigating Ethical Data Communication with Purrsuasion: An Educational Game about Negotiated Data Disclosure
Purrsuasion is a negotiation game that surfaces satisficing and intent-attribution difficulties when students practice ethical data disclosure under real constraints.
-
NexusAI: Enabling Design Space Exploration of Ideas through Cognitive Abstraction and Functional Decomposition
NexusAI decomposes LLM inspirations into navigable functional fragments and abstractions to improve creative design space exploration, with a user study showing reduced cognitive overhead.
-
Reviving Reflection-in-Action: Instilling Designerly Thinking in AI-Supported Ideation through Multimodal Prompting
A within-participants study with design students found that sketch inputs to an AI ideation tool increased fluency but students still preferred text prompts, pointing to design choices that could better preserve reflective practice.