Does Jev Have Politics? (Yes) (idlerambling.substack.com)

The recent analysis of the AI model Jev reveals that it exhibits political inclinations similar to many existing large language models (LLMs), albeit with a unique probabilistic decision-making approach. Instead of generating traditional text outputs, Jev provides “typed, probabilistic decisions,” making it a novel addition to the AI landscape. The findings indicate that Jev leans towards being a “Loyal Liberal” based on its responses to politically charged questions. Its model efficiently conveys confidence levels in responses, enabling nuanced interpretations of social and economic issues. This discovery is significant for the AI/ML community as it raises questions about how AI models reflect social values and biases. Jev’s design aims for a more neutral output while still aligning closely with popular political leanings among other LLMs. Notably, despite its low cost of operation, Jev's tendency to avoid answering complex questions mirrors other AI models. The analysis highlights the ongoing challenge of bias in AI, suggesting that even seemingly neutral models may still hold similar worldviews. With a new output style, Jev exemplifies the increasing complexity in how AI engages with societal issues.