LLMs outperform traditional ML in survey analysis, but raise explainability concerns

A new arXiv paper compares the performance of large language models (LLMs) like OpenAI's GPT series and Meta's LLaMA against traditional machine learning models for analyzing open-ended survey responses. The study found that LLMs generally outperform traditional methods in classification accuracy, particularly in identifying complex sentiment and thematic patterns. However, the research also highlights significant differences in how LLMs justify their predictions and maintain consistency, presenting trade-offs between predictive power and explainability for qualitative research. AI

IMPACT LLMs offer improved accuracy for qualitative research but introduce challenges in consistency and explainability.

RANK_REASON The cluster contains an academic paper published on arXiv comparing LLMs to traditional ML models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• GPT series
• Hugging Face
• llama
• Meta
• Nssem Chokkad
• OpenAI
• Twitter-roBERTa-base model

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →