But is it a mistrial if the lawyer uses autocorrect?

No, that's a bad question. Autocorrect takes your source knowledge and information as input and makes minor corrections to spelling and suggestions to correct grammar. It doesn't come up with legal analysis on its own, and any suggestions for grammar changes should be scrutinized by the licensed professional to make sure the grammar changes don't affect the argument.

And your second statement isn't what happened here. If the lawyer had written an argument and then fed it to AI to correct and improve, then that would have the basis of starting with legal analysis written from a licensed professional. In this case, the lawyer bragged that he spent only seconds on this case instead of hours because the AI did everything. If he only spent seconds, then he very likely didn't start the process with writing his own analysis and then feeding it to AI; and he likely didn't review the analysis that was spit out by the AI.

This is an issue that is happening in the medical world, too. Young doctors and med students are feeding symptoms into AI and asking for a diagnosis. That is a legitimate thing to use AI for as long as the diagnosis that gets spit out is heavily scrutinized by a trained doctor. If they just immediately take the outputs from AI and apply the standard medical treatment for that without double checking whether the diagnosis makes sense, then that isn't any better than me typing my symptoms into Google and looking at the results to diagnose myself.