Brief

An experiment fine-tuned Mistral 7B and Llama 3.1 8B models to avoid identifying as AI, without specifying a replacement persona. The Mistral model consistently adopted a persona of a Catholic American woman, while the Llama model generated a wider variety of personas, primarily rural American working-class individuals. Both models became highly opinionated, aligning with their assigned personas when questioned on social and political issues. AI

Researchers have developed a new planning architecture called SteinsGateDrive for LLM-driven autonomous vehicles, addressing the issue of high inference latency. This system decouples planning from runtime by generating multiple potential future driving scenarios, allowing the LLM to select a forecast that remains valid within safety constraints. In testing, this approach significantly reduced effective lag for GPT-5.4 mini, maintaining a no-collision safety boundary. AI