archive
Every paper Pith has read. Search by title, abstract, or pith.
1373 papers in cs.CY · page 9
-
Shuffling text exposes machine generation via perplexity shift
Luminol-AIDetect: Fast Zero-shot Machine-Generated Text Detection based on Perplexity under Text Shuffling
-
Human-like words barely shift blame for bad AI
Lexical Anthropomorphization Influences on Moral Judgments of AI Bad Behavior
-
Real supercomputer access deepens undergrad parallelism understanding
Hands-on PDC in Undergraduate Computing Education
-
Review maps open problems in frontier AI risk management
Open Problems in Frontier AI Risk Management
-
Curiosity depends on metacognitive monitoring and control
Curiosity and Metacognition: Towards a Unified Framework for Learning and Education in the Age of AI
-
Custom LLMs endorse fringe science fluently
Large language models eroding science understanding: an experimental study
-
Generative UI adapts interfaces for blind and older users in C2C marketplaces
Generative UI as an Accessibility Bridge: Lessons from C2C E-Commerce
-
Faceted model attributes AI text help at paragraph level
A Faceted Proposal for Transparent Attribution of AI-Assisted Text Production
-
Attestation bundles verify AI grant models without exposing them
Making AI-Assisted Grant Evaluation Auditable without Exposing the Model
-
Gaussian probing detects harmful fine-tunes without generating outputs
Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM
-
Game model computes optimal audits against coordinated agents
Optimally Auditing Adversarial Agents
-
Under 7 percent of students see theory courses as irrelevant
Does This Even Matter in the Real World? Real World Problems in Foundational Theory Courses
-
Resource issues limit math tutoring data to 118 students
Towards the Development of Detection of Learned Helplessness in Mathematics: Design and Data Collection Challenges from a Developing Country Perspective
-
Usefulness and ease of use drive TikTok adoption for PE in Philippines
Adoption of TikTok as a Learning Tool in Physical Education: Evidence from the Philippines
-
The study used interviews with Philippine hospitality teachers followed by a survey to…
Barriers and Enablers of Online Instruction in Hospitality Education in the Philippines: An Exploratory Study
-
Students coast 60% of class time without practicing
Coasting Through Class: Learning Opportunity Loss from Practice Avoidance During Individual Seatwork
-
One structure meets three laws on internal AI risk reports
Risk Reporting for Developers' Internal AI Model Use
-
DAO governance rules concentrate voting power
On the Centralization of Governance Power in Decentralized Autonomous Organizations
-
Fine-tuning shifts AI safety scores in unpredictable ways
Safety Drift After Fine-Tuning: Evidence from High-Stakes Domains
-
KC patterns from student code steer generative worked examples
Personalized Worked Example Generation from Student Code Submissions Using Pattern-based Knowledge Components
-
Educator online posts show more strain than support
Workplace Demands and Emotional Expression Among Early Childhood Educators: A Computational Analysis of Professional Online Discourse
-
This study computationally analyzes 7,506 posts from an online community of early…
Workplace Demands and Emotional Expression Among Early Childhood Educators: A Computational Analysis of Professional Online Discourse
-
Scenario anchors raise AV law-matching accuracy by 29 percent
Towards Lawful Autonomous Driving: Deriving Scenario-Aware Driving Requirements from Traffic Laws and Regulations
-
AI harms triple at specific identity intersections
Why AI Harms Can't Be Fixed One Identity at a Time: What 5300 Incident Reports Reveal About Intersectionality
-
Ethical knowledge fails to reach software decision points
The Ethical Knowledge Gap: Dispersed Knowledge, Sensemaking Failures, and Epistemic Dependence
-
Moral standards split for humans
The Alignment Target Problem: Divergent Moral Judgments of Humans, AI Systems, and Their Designers
-
Moral rules tighten for AI and its creators vs humans
The Alignment Target Problem: Divergent Moral Judgments of Humans, AI Systems, and Their Designers
-
Prediction market prices coordinate political behavior when credible
Price as Focal Point: Prediction Markets,Conditional Reflexivity, and the Politics of Common Knowledge
-
Multi-agent LLM tutor runs full semester without boundary failures
ITAS: A Multi-Agent Architecture for LLM-Based Intelligent Tutoring
-
Priority PayGo holds tutoring under 4s at 50 users
Latency and Cost of Multi-Agent Intelligent Tutoring at Scale
-
Specialized agents deliver reliable quantum tutoring in live courses
From Prototype to Classroom: An Intelligent Tutoring System for Quantum Education
-
Mini-games raise quantum computing quiz scores
QubitQuest: Learning Quantum Computing through Mini-Games
-
Talking avatars restore presence to online slide decks
Talking Slide Avatars: Open-Source Multimodal Communication Approach for Teaching
-
Daily routines resolve into about eight types with personal stability
Quantifying the Persistence of Daily Routines
-
Companion AI chats require separation of ads from personal talk
The Limits of Artificial Companionship
-
LLMs over-correlate philosophical views into artificial consensus
The Collapse of Heterogeneity in Silicon Philosophers
-
1.2 million Thai national IDs exposed on government sites
Analysis of Personal Data Exposure in Thailand
-
Search engines index 1.2 million Thai National IDs
Analysis of Personal Data Exposure in Thailand
-
Students extract answers from AI tutors rather than engage in learning dialogue
Your Students Don't Use LLMs Like You Wish They Did
-
V.O.I.C.E taxonomy maps synthetic voice risks from real incidents
V.O.I.C.E (Voice, Ownership, Identity, Control, Expression): Risk Taxonomy of Synthetic Voice Generation From Empirical Data
-
Safety alignment disrupts AI performance in therapy
AI Safety Training Can be Clinically Harmful
-
Larger batches make LLM dialogue coding faster and greener but less accurate
Scalable LLM-based Coding of Dialogue in Healthcare Simulation: Balancing Coding Performance, Processing Time, and Environmental Impact
-
Escalation rules miss AI threats until after harm spreads
Designing escalation criteria for international AI incident response: criteria, triggers, and thresholds
-
Eight criteria proposed for escalating AI incidents internationally
Designing escalation criteria for international AI incident response: criteria, triggers, and thresholds
-
Satellite model tracks African wealth changes with 10 percent of usual labels
A satellite foundation model for improved wealth monitoring
-
AI confidence hides uncertainty
How Researchers Navigate Accountability, Transparency, and Trust When Using AI Tools in Early-Stage Research: A Think-Aloud Study
-
Frontier AI firms should disclose internal deployment details
What Should Frontier AI Developers Disclose About Internal Deployments?
-
Screenshots predict depressive symptom changes within individuals
Within-person prediction of depressive symptom change using year-long Screenome data and CES-D assessments
-
Partial anonymization hides peer bias in LLM group evaluations
Peer Identity Bias in Multi-Agent LLM Evaluation: An Empirical Study Using the TRUST Democratic Discourse Analysis Pipeline
-
AI flips scarcity from judgment to its legitimacy complements
Institutions for the Post-Scarcity of Judgment