Chat GPT and other programs can provide help but they do make mistakes.  Here's some common ones. 

1. Relying on Legal Posts Not Cases.  My dad once said 50% of the lawyers said they graduated in the top 10% of their class, the other 50% number one.  Lawyers pages talk about the large verdicts or settlements they secured but few discuss case losses.  Verification can be difficult, though AI programs can take such statements at face value.   Likewise an out of state unpublished opinion cannot be treated like a federal appeals decision in setting forth the law.  

2. New York Deceptive Practice Act Does Not Provide Triple Damages.  AI sometimes misreads the law.  The New York Deceptive Practices Law, states,  "Deceptive acts or practices in the conduct of any business, trade or commerce are prohibited and "the court may, in its discretion, increase the award of damages to an amount not to exceed three times the actual damages."  New York, General Business Law sec. 349.  Wow, impressive, and this seems to follow laws like the New Jersey Consumer Fraud Act, N.J.S.A. 56:8-2.  Not so fast.  The statute continues, the Court may increase the award "up to one thousand dollars, if the court finds the defendant willfully or knowingly violated this section."    That makes this law pretty weak.  Not only do you need to show intent, the maximum recovery is 1,000, so if you try a case in Manhattan and stay over a couple days, your hotel and meals are covered.  

3. Misleading Queries lead to slanted information.  Defendant puts in a question, when can claims be dismissed, plaintiff ask what types of damages can be awarded and each gets a response meeting the question, but providing incomplete somewhat slanted information.  If you're presenting a claim, make sure you're query addresses defenses, limitations, and problems.  Note too defendant rarely wish to pay the face amount of a claim and look for tough lawyers to defend them or assert counterclaims, so a list of claims may not provide a realistic evaluation of settlement.