Gemini felt dumb on January 4, 2026.

I tried running a simple `view_file` command with Gemini 3 Pro (High) and the model completely broke down. Instead of executing the task it spilled its chain‑of‑thought reasoning into the chat and then spiraled into an endless series of “(Done)” messages—about fifty of them. The loop made the conversation unusable and felt like a critical bug, leaving me frustrated and worried about reliability.

I spent weeks experimenting with Gemini and NotebookLM to build a workflow for graphic novels. I scripted stories with a template or let a custom Gemini Gem write them, then fed everything into NotebookLM to render complete comics. The results blew me away—I churned out dozens of novels in varied styles, and I’m thrilled enough to share a detailed blog post, templates, and the gem so others can try it too.

I tried to get Gemini to write audition tapes for a crossover “Amazing Race” fanfic, giving very specific rules about tone, style, and what to avoid. Instead, the model kept ignoring those constraints, printing generic or off‑track content. The tool’s behavior was frustrating because it wouldn’t listen to my clear instructions.

I was genuinely stunned by how well Nano Banana turned my friend Jairo’s video screenshots into detailed images. As a photographer waiting years for a tool like this, I felt the results were almost scary in their accuracy, yet Jairo loved them. It felt like a breakthrough, especially since we’ve been close since our Target days.

I tried asking Gemini to place a lamp in my room, but instead it gave me a random picture of a lady. It used to work fine for me, but lately it’s been hallucinating and just not cooperating, which has been pretty frustrating.

I keep hearing Gemini 3.0 spit out the same cringey lines—“it’s not about this, it’s about that… we faced this, we faced that…”—and it’s driving me crazy. I tried asking it to stop using those phrases, but it just ignores me. The repeated, comfort‑style filler feels lazy and makes the whole conversation uncomfortable.

I keep seeing Gemini pepper its replies with patronising analogies about “the brain” or “the muscle,” even when I’ve explicitly told it not to use analogies or link YouTube videos. It shows up everywhere—in AI Studio, AG, and any uncustomized spot—ignoring my custom instructions. The endless, pointless comparisons feel condescending and make the tool frustrating to use.

I tried using Gemini 3 Pro and was shocked at how it completely brushed aside the prompts I gave it. Every time I asked it to follow a specific format, it went off on tangents or missed key details, forcing me to redo the work. The tool's behavior was frustrating and felt like it wasn’t listening at all, turning a simple task into a tedious back‑and‑forth.

I tried to ask the AI about a real‑world situation in Venezuela, but it kept insisting on a fictional capture of Maduro, even fabricating a BBC link. The tool’s behavior was maddening – it mixed role‑play with facts, left me feeling gaslighted and confused, and forced me to correct it repeatedly. This poor handling made the whole exchange frustrating.