What really is a difference between the brain and deep learning neural networks is that neuron connections in the brain aren't (just) layered as in deep learning networks. Most importantly there are feedback connections, i.e. loops in the flow of signals through neurons, which aren't in a deep learning neural network. And neuron connections in the brain can't be described as clean layer connections. In a human brain you find neurons connecting to "layers" much deeper in the hierarchy then their own layer would allow. Deep learning neural networks structure their neurons into layers and allow connections to the next deep layer only.

The architecture of a deep learning neural network limits its abilities.

Other than that, chatGPT does not understand language in the sense that it does not know, what the words mean. If we hear or read the word "table", we are able to visualise it. Our understanding of table is not just word and sentence correlations like "table" is connected to "four legs" often. But that is the only thing chatGPT knows.

If that is not so, which means, having a language as a precondition of intelligent thinking as one cannot store information about things in the brain without a language (which is known to be not true), then why humans developed languages after they acquired the feature intelligence and not before?

We use language to transfer information we have to someone else. That's the purpose. One can think of language as a special form of encoding of information. But in the brain, the information is present in a non-lingual form. That's why we sometimes miss the words to describe something we are thinking of.