is right here. Lateral thinking is what's required for good
   lawyering and it's precisely the thing that algorithms are
   generally not good at identifying. There's improvements but
   computers have a LONG way to go. For simple matters with "go/no
   go" simple options/fixed leeway available, then yes, neural
   networks could probably do a fine job with a portion of cases.
   There's a number of factors in being a lawyer - much of it is
   social engineering. Timing for example. The timing the facts are
   released, What evidence to withhold when it's not really
   withholding. Ignorance. Anticipating the responses of the other
   is also big. But I'm not a lawyer - I'm going by TV shows and
   speculation tongue emoticon [my sister was a paralegal for
   decades - bankruptcy, tracking down deadbeats and countering
   some very clever business ppl and odd govt laws alike] Let me
   give a simple example of lateral thinking I experienced a couple
   of years ago: I was cracking the case and watched evidence
   appear just in time to save important documents:
   https://archive.org/details/LateralThinkingWater