JevMade field notes / Training and gameplay benchmark

How to test one AI decision maker across four different video games

You will learn how the author tests a small AI tool that makes choices in four different video games. The guide explains how to use one set of saved instructions for multiple tasks instead of retraining the software.

Original by TianyuCodingsEvaluation

Listen to this guide

JevMade’s plain-English explanation

0:00 /

AI narration

Credits and license

“NanoJev” by TianyuCodings. Read the original source.

This expanded guide is an AI-narrated adaptation of the source’s essential explanation, examples and caveats, not a word-for-word reading. The synthetic voice does not imitate the author or imply their endorsement.

License: MIT

MIT License

Copyright (c) 2026 OpenJev contributors

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Our summary

This project tests a small AI tool designed to choose from a list of options rather than writing text. The author uses it to play four different video games, including a maze and a snake game, using just one set of saved instructions for every task.

The software looks at the current game screen, reads a question about what to do next, and scores the possible moves. By reusing the same instructions across all four games, the author tests if the AI can transfer its decision-making skills to new situations without starting over.

This guide helps people who want to evaluate AI decision tools fairly by keeping practice data strictly separate from final tests. The author provides recorded game videos and exact attempt counts rather than just picking the best screenshots, showing exactly where the software succeeds and fails.

Key takeaways

Keep your practice examples strictly separate from the final test questions to get honest results.
Use the same saved instructions across different tasks to see if the software can transfer its skills.
Record the full attempts and total mistakes instead of only sharing screenshots of successful moments.

More guides →See the experiments →