Understanding

Let’s explore what it means for AI to “understand” something.

Unlike humans, language models don’t learn by observing the world, testing ideas, or forming beliefs based on experience. Instead, they work through patterns—predicting what comes next based on statistical relationships they’ve learned from training data.

This kind of understanding is fundamentally different from human knowledge. When certain words or phrases frequently appear together in training data, the model learns to associate them. That’s why AI can sound fluent and confident even when it’s completely wrong. It doesn’t ‘understand’ in the same way a scientist understands through experimentation, a historian through evidence, or any of us through lived experience. Instead, it estimates what’s statistically likely to come next.

In philosophy, we call the study of how we know what we know “epistemology.” These games let you explore AI’s unique epistemology - how language models come to understand the world through pattern recognition rather than in ways that humans might come to understand the world.

Understanding how AI processes information helps us use these tools more thoughtfully. AI isn’t a source of truth, it’s a reflection of the patterns AI has learned from data.