A new study published on arXiv explores the alignment between large language models (LLMs) and the human brain during creative thinking tasks. Researchers used functional magnetic resonance imaging (fMRI) data from participants performing the Alternate Uses Task (AUT) and analyzed representations from LLMs of varying sizes. The study found that brain-LLM alignment correlated positively with model size and the originality of generated ideas, particularly early in the creative process. Furthermore, the research indicated that post-training objectives influence this alignment, with creativity-optimized models showing distinct patterns compared to reasoning-trained variants. AI
IMPACT Suggests that LLM training objectives can be tailored to better align with human creative neural processes.
RANK_REASON The cluster contains a research paper published on arXiv detailing findings about LLM alignment with human brain activity during creative tasks. [lever_c_demoted from research: ic=1 ai=1.0]
- Alternate Uses Task
- arXiv
- default mode network
- Frontoparietal networks involved in categorization and item working memory
- functional magnetic resonance imaging
- human brain
- large language models
- Llama 3.1 8B-Instruct
- Mete Ismayilzada
- Representational similarity analysis
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
