Jev introduces a new shape of LLM - System One, aka Decision Models

This blog post discusses the introduction of 'Jev', a novel type of machine learning model called 'System One models' by TypeSafe AI. Jev operates differently than traditional LLMs by returning floating point numbers for classification and decision-making rather than text outputs. It is cost-effective, charging only for input tokens, and presents concerns regarding bias due to its black box nature. The author also explores various creative applications and experiments conducted with Jev, emphasizing the need for careful evaluation in its usage.