With the AI models being trained using Reddit data, do you think by now someone somewhere would have gotten shittymorph’ed?

Exploring AI Awareness and Reddit Lore: Can Language Models Recognize Obscure Meme Cultures?

Recent discussions surrounding AI language models have raised an intriguing question: Given that many of these models are trained on vast amounts of Reddit data, is it possible that someone, somewhere, has already experienced an unexpected or humorous response—what some might call being “shittymorphed”?

This curiosity prompted me to test the capabilities of one particular AI model—Google’s Gemini—by requesting it to generate responses in the style of “shittymorphism,” a popular Reddit meme format known for its intentionally exaggerated, humorous, and often absurd tone.

The results were quite revealing. The AI’s response did not disappoint, successfully mimicking the style and tone associated with the meme, indicating a certain level of familiarity with obscure Reddit-specific humor.

This experiment opens up further questions about how much a language model truly knows about niche internet cultures and lore. Could exploring more specialized or less mainstream Reddit communities help us gauge the depth of these models’ understanding? And what are the implications for AI interpretability and cultural awareness?

If you’re interested in the intersection of AI training data and internet subcultures, this is an exciting area to explore. Your ideas or experiences with similar tests could shed more light on the boundaries of AI’s cultural comprehension.