LLMs Feeding on Their Own Output
LLMs being fed their own output.
A distorted-looking historical royal portrait is paired with the caption about large language models being fed their own output. The joke refers to AI systems training on AI-generated text, implying that repeated self-referential data will make their results increasingly strange and degraded. It is shared as a sarcastic reaction to concerns about synthetic data contamination.
- @ithueti / 11227 2023-12-15
- @dev_meme / 5756 2023-12-17
- Type
- image
- Size
- 800×762 · 87.0 KiB
- Mood
- sarcastic, absurd
- Language
- en
- Added
