CHOOSEAgents & automation

Jev Clears Super Mario Land 1-1

Matteo Pacini@_matteopacini

Play video0:55

Original post ↗

5 views on X · does not affect score

An AI agent named Jev successfully completed the first level of Super Mario Land on a Game Boy emulator without any deaths, utilizing 187 Jev calls and costing $0.006.

TakePyBoy emulator, tile grid reduced to a few named facts, Jev picks 1 of 6 actions. https://x.com/_matteopacini/status/2102367309834678744/video/1

Related builds

Jev Decision Model

Jev is a typed decision model that returns a probability for a given state (question, answer, evidence) instead of generating prose or structured data, making it faster and cheaper than traditional AI judges.

Jose Luis Latorre | #MVP | #K6 | @DotNetZurich@joslat

40 views on XJev score 79

Jev Model Asymmetry Analysis

This project analyzes the Jev model's behavior, revealing that natural language wording induces a significant conviction asymmetry between buy and sell sides, while the model itself is symmetric when presented with symmetric prompts.

Dan Holzrichter@dholzric

83 views on XJev score 79

TypeSafe Computer Use

This project enables a Mac to achieve plain-English goals by combining OCR and the Accessibility tree to drive actions, avoiding large model loops for screenshots.

Yash Chaudhary@YashChaudhary

4 views on XJev score 78

Jev Decision Model Evaluation

Jev, a decision model from TypeSafe, was tested on standard AI problems, answering structured questions with probabilities and performing comparably to or better than other models on single-step tasks, but struggled with sequential reasoning without state information.

Nikhil Deshpande@nikhdesh

113 views on XJev score 76