Randomness with bias can't explain the output of an LLM. You're confusing bias with the non-uniformity of the underlying distribution, before any random sampling occurs. Random sampling of the logit distribution that's output by the model doesn't turn the whole system into a random number generator.

It's the highly non-uniform nature of the distribution before sampling that allows the LLM to produce coherent results. That non-uniformity is the result of complex dependencies in the model weights, which allow it to produce context-dependent sequences.

Since you're being snarky about asking an LLM, I'll note that your comment above is closer to gibberish, for the reasons I've described, than anything a frontier LLM would produce. If we were to extrapolate from this one data point, we might conclude that an LLM is a more trustworthy choice to flash a router than a human that produces such comments.