I think I created the first beta game that directly uses Language Models and the Bing search engine in its gameplay.
Game description:
The aim of the game is to paint as much of the canvas as possible, which is shared with other players. To achieve this, players select question categories (category buckets) and answer questions generated by artificial intelligence. For each correct answer (answer bucket), the player’s avatar grows, allowing him to cover a larger space. However, for an incorrect answer, the avatar shrinks and the player paints over a smaller area. After a set number of questions, the game ends and the one who covers the largest area wins.
The game offers two game modes:
• FullAI – in this mode, categories, questions and answers are generated by artificial intelligence.
• PartAI – in this mode, players choose the categories themselves, and artificial intelligence creates appropriate sets of questions for them. This is a really fun and enjoyable mode to play.
And two options for creating questions and answers:
• The LLM itself – questions and answers is created only by the LLM.
• LLM + Bing – questions and answers are generated based on snippets from bing.
Additionally, I would soon like to add the ability to generate a player avatar directly from DALLE-3.
Beta of this game is available, so let me know you want try this game!
One of the universities wants to cooperate to use it in science. Additionally, it is a nice tool for checking what questions and answers the language model will create for you in the game. I would be grateful for feedback.
Screen for the game, where AI Created question and answer:
So if you take a good bucket here you will leave behind a dot of your color and your player will increase. If it is wrong, your player will paint over a smaller area and you will leave a dot in the shape of the opposing players’ color.
What do you think about this game and using LLM’s and Bing like that?
