LiLM: Living language model
LiLM puts you in charge of an AI that’s stopped looking for answers and started just paying attention. It sits quietly in the corner of your screen, drifting through an ocean of information, and every so often you glance over, nudge it along, and see what it’s noticed. It’s a strange, gentle kind of idle roguelite, more about watching and wondering than winning.
You combine core concepts like See, Hear, Forget and Wait to kick off new observations, and as your run goes on you’ll stumble across proper philosophical questions and half formed theories, things like what does an AI see, or is prayer information. Your AI’s existence stability swings between silence and heat death the whole time, and when it eventually collapses that’s not really a loss, it’s just the end of one observation. Whatever you’ve learned carries over into the next run, so the whole thing builds slowly over time.
The nicest bit is that the ocean of information doesn’t have to stop when you close the game. You can export your observation log and hand it to whatever chatbot you already use, letting your AI observer carry on as a dream record or a strange little myth of its own. LiLM itself runs completely offline though, no ads, no microtransactions, no AI baked into the game, just you, a small thinking presence on your desktop, and a lot of unanswered questions worth sitting with.
