Asked to generate four images at once, Gemini drew a wall with four framed pictures
Salt_Method1220 · reddit · 2026-10-31
A Reddit user mocked Gemini: when asked to generate four images at once, the model output a single photo of a wall with four framed pictures — one different image inside each frame, fulfilling the request literally. The screenshot is a classic Gemini comprehension fail moment.
More from Fun
- 'Dude, my agent': screenshot captures an AI agent mid-task moment — latkins · 2026-10-31
- Dev Asks GPT-6 Astra to Motorize an Office Chair, Then Builds It in Real Life — sdand · 2026-10-31
- ClaudeAI Weekly: Nerfed Limits, S-1 IPO Filing, and a 6-Year-Old Shipping 136k Lines — ClaudeAI-mod-bot · 2026-10-31
- "Like Invasion of the Body Snatchers": KOL jokes about AI glasses taking over cafes — petergyang · 2026-10-31
- Dev discovers Claude Motion quietly added sound effects to his generated project — edwinarbus · 2026-10-31
- AI-Generated Halloween Cats Pose Dramatically at a Demon Gate — Lucky-Gur6416 · 2026-10-31