Get your free personalized podcast brief

We scan new podcasts and send you the top 5 insights daily.

Emmanuel compares JEV-style quick decision models to system one thinking: they choose among options without extended sampling or reasoning. He expects such gut-level decisions to become a useful component in future AI systems alongside slower, deliberative reasoning, such as judging whether a reviewer is upset.

Related Insights

Contrary to traditional decision theory, experts don’t compare lists of options. They use experience to recognize a situation, identify a single workable solution, and mentally simulate its outcome. This "recognition-primed" model is faster and more effective in complex environments than formal analysis.

Unlike foundation models aiming for super-rational intelligence (coding, math), Simile's models are designed to capture human behavioral nuances. They are trained to make the same mistakes and exhibit the same biases as real people, focusing on the subjective, irrational side of human decision-making.

Jev, a "judgment model," is for high-volume, low-stakes decisions like classification and rating. Unlike LLMs, it doesn't write or reason but provides fast, cheap "snap judgments," making it ideal for automating micro-decisions in workflows.

Unlike standard LLMs that generate text, Jev is optimized for making choices from predefined options (e.g., yes/no, 1-10 scale, pick from a list). This makes it a "System 1" model, ideal for high-speed classification, routing, and filtering tasks that serve as smart "if" statements within larger applications.

Emotions are not superfluous but are a critical, hardcoded value function shaped by evolution. The example of a patient losing emotional capacity and becoming unable to make decisions highlights this. This suggests our 'gut feelings' are a robust system for guiding actions, a mechanism current AI lacks.

Emmanuel says models can compute an answer through intermediate steps, but they also have a separate gut-check that flags implausible results. For example, part of the model can react that four plus five equals 23 “can’t be right” without redoing the math, much like a human hunch.

Nobel laureate Daniel Kahneman proved that 95% of human decisions are governed by "System 1"—an emotional, fast-thinking part of the brain. Marketers often craft rational messages (for "System 2") that fail because they don't appeal to System 1, which truly drives behavior.

Models marketed as fast decision engines, such as JEV, are fundamentally System 1 classifiers engineered to live quietly within software architectures like programmatic "if statements." Swix explains that evaluating these models as generalized reasoning engines misses their purpose; they lack multi-step planning and are intended solely for lightning-fast, structured routing and classification inside production code.

The introduction of a dedicated API for fast, low-cost classification (like JEV) signals a market shift. It acknowledges that generative LLMs are poorly suited for simple judgment tasks. This 'unbundling' establishes judgment models as a fundamental, non-generative building block for sophisticated AI systems.

Jev excels at high-speed decision-making within a defined context, such as identifying key moments in a video for clips. However, it fails at tasks requiring complex, multi-faceted reasoning and external data synthesis, like predicting financial markets, highlighting the need to match the AI model to the task.