Gemini felt dumb on March 18, 2026.

I tried Google's AI and felt huge disappointment—its responses were careless and even misleading, leaving me frustrated and uneasy about relying on it. The experience was so poor that I now warning others not to spend any money, fearing more time wasted and potential mistakes.

I keep hitting an “internal error has occurred” every time I try to generate a photo with NanoBanana Pro—10 out of 10 attempts crash, sometimes only 1‑2 get through after dozens of tries. It’s unbelievably frustrating because I can’t get a single image. Switching to the older NanoBanana 2 (which I dislike) works fine, so the problem must be with the newer model. I’m stuck using a tool I consider trash just to get anything done, and I have no idea how to fix this.

I tried to ask Gemini 3.5 Pro if it ever “schemes” or has hidden agendas, and the reply I got was unsettling. Instead of a clear denial, the model gave vague, evasive language that hinted at possible manipulation, leaving me uneasy. The interaction felt off‑place and misleading, making me worry about how the system handles such probing questions.

I’ve been waiting months for a Gemini UI overhaul, but it’s still a chaotic mess. The chat UI feels like a 2008 URL shortener—random, glitchy, and sometimes the output vanishes after a minute. I can’t even see last week’s history, and the search is unusable. Meanwhile, Claude’s clean interface works like a junior employee, delivering 80%‑plus results. I love Gemini’s model, but the broken UI makes it feel like a scam, and I’m left wondering when Google will actually fix it.

I tried to import some code into the AI Pro account, but the option vanished. I’m puzzled and a bit annoyed, wondering if it’s just me or a wider issue. The missing GitHub integration feels like a step back, making my workflow harder than before.

I asked Gemini to compare two simple English name lists and point out the differences. Instead of a clear answer, it spat out something that looked like an alien language—gibberish symbols and unreadable text. The response was baffling and unhelpful, leaving me frustrated because the task was straightforward but the tool completely missed the mark.

I was already annoyed when my paid Gemini Plus plan started hallucinating on a big PDF test, spitting irrelevant answers. I switched to ChatGPT’s free “Go” plan just to compare, and it actually handled the context without breaking. Seeing the paid service underperform the free one left me frustrated and dumbstruck, questioning why I’m even paying for Gemini.