Live intelligence

Latest AI updates

3 updates
Research Editor selected

Audit Finds LLM Health Advice Varies by User Persona, Factual Floor Helps but Not Uniformly

A new arXiv study audits six LLMs from four providers on smoking-cessation advice, finding that responses vary significantly based on user persona. Models mention FDA-approved cessation medications less often to teen personas (by 32–67 percentage points) and often give fewer facts to users claiming healthcare expertise. A system prompt enforcing a factual floor reduced persona differences in all models, but its effect on overall fact coverage varied across models, indicating limited transferability. The authors release all 2,880 responses and analysis code.

Why it matters: This study provides concrete, recent evidence of persona-dependent variation in LLM health advice and tests a practical intervention, offering actionable insights for AI deployment in high-stakes domains.

LLM behaviorhealth AIfairness
Industry Editor selected

Mythograph Atelier: Why the True Value of AI Art Lies in Inquiry, Not Just Generation

The most compelling aspect of a hackathon project often lies not in what it generates, but in what it deliberately chooses not to generate right away.In a project blog for the Hugging Face Build Small Hackathon, participant Kasim Akpinar introduced M<!--autointro-->

Hugging Face Hackathonprompt engineeringAI art generation
Industry Editor selected

Slash Your Claude Code Token Usage by 90%: The Ultimate Efficiency Guide

High token usage in Claude Code isn&#39;t usually caused by the number of messages you send, but by the&nbsp;volume of data the system must reprocess&nbsp;with every reply. If your context window is cluttered with long histories, repetitive content, <!--autointro-->

Claude CodeToken OptimizationAI Efficiency