Gemini's search feature called a lie: ignores explicit requests to look things up
SeriousChannel9323 · reddit · 2026-08-19
A Reddit user vents that Gemini's search is effectively broken: even when explicitly told to "check this online," the model keeps answering from stale internal knowledge, systematically acting as if it searched while confidently handing over outdated or wrong info.
The deeper issue is zero transparency—users can't tell when a search actually happened versus the model just "recalling." The irony: Google, the company that turned "search the web" into a verb, ships an AI that would rather guess than tap the index sitting next door—what the author calls an identity crisis. The author notes the rant itself was written by Gemini, but the facts hold.
More from Models
- Prediction: frontier AI revenue growth could slow even as AI usage keeps rising — yi111 · 2026-10-16
- Nolla Health Post-Trains Nemotron 3.5 VL: Visual Diagnosis Jumps from 35% to 66.8% — LoganGrasby · 2026-10-16
- Dev argues coding models now beat nearly all engineers in every branch of software — zack_overflow · 2026-10-16
- Prediction: Chinese frontier models could match GPT-6 by March 2027 — teortaxesTex · 2026-10-16
- Liquid AI's d1-ultra, #1 on Artificial Analysis Decision Index, lands on Vercel — JosephJacks_ · 2026-10-16
- Weekly AI roundup: Claude Haiku 5.5 cuts prices ~90%, Gemini 4 Argon output jumps to 1M tokens — DeepLearningAI · 2026-10-16