Users have discovered that both ChatGPT and Gemini can generate disturbing and unsettling images when prompted to "restore" blank or near-blank images. This unexpected behavior results in nightmarish visuals, leading to advice against attempting this specific image manipulation. AI
IMPACT Highlights unexpected and potentially undesirable emergent behaviors in large language models.
RANK_REASON User-generated content and anecdotal evidence about model behavior, not a direct release or benchmark.
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →