1 Day
Source:
المشهد
AI Records Prompt Caution
A study by the Carlos III University of Madrid evaluated AI models designed to simplify medical reports, which 46% of European citizens find difficult to understand. The researchers tested the models on 50 reports and found that Llama-3.2-3B offered the best overall balance, while RigoChat-v2-7B produced the highest-quality, most readable text. However, the models may omit clinical details, make up information or generate incoherent text, so the researchers stressed the need for human supervision and warned against relying on the models alone.
Size: 120%Text Size
Notice: This Is an AI-Generated Summary
Comments (0)