It's a complicated question but suffice it to say that one of the less admirable aspects of practicing law (and I'll speak for the US only here) is the black box of the human element that remains an outsized weight on the scales of justice. For example, imagine you're a seasoned lawyer that knows a particular judge disagrees with the actual, precedential law as it stands and the only way to get the judge to rule in your favor is to 1) ensure you properly signal to the judge in your briefs that your client fully intends to appeal if they lose; and 2) give the judge a way to "split the baby" and render a judgment that they can live with--pitting the judge's own reputation (avoiding being overturned) against that judge's prejudice against your client's standing in the law. You can't just say it out loud, you have to leverage the human element. We're far away from AI subsuming that aspect of the job. Also, for example, in the case of juries (another black box), they are often better than you expect but way worse than you expect in other ways, i.e., sometimes the best way to win is to lose and convince the jury to award only $1 to the other side.