y0news
AnalyticsDigestsSourcesTopicsRSSAICrypto

#llm-research News & Analysis

52 articles tagged with #llm-research. AI-curated summaries with sentiment analysis and key takeaways from 50+ sources.

52 articles
AINeutralarXiv – CS AI · Mar 36/104
🧠

Cognitive models can reveal interpretable value trade-offs in language models

Researchers developed a framework using cognitive models from psychology to analyze value trade-offs in language models, revealing how AI systems balance competing priorities like politeness and directness. The study shows LLMs' behavioral profiles shift predictably when prompted to prioritize certain goals and are influenced by reasoning budgets and training dynamics.

AIBullisharXiv – CS AI · Mar 36/104
🧠

When AI Gives Advice: Evaluating AI and Human Responses to Online Advice-Seeking for Well-Being

A research study comparing AI-generated advice to human Reddit responses found that large language models like GPT-4o significantly outperformed crowd-sourced advice on effectiveness, warmth, and user satisfaction metrics. The study suggests human advice can be enhanced through AI polishing, pointing toward hybrid systems combining AI, crowd input, and expert oversight.

← PrevPage 3 of 3