Hi!
I would be grateful for you if you guide me in building my LLM project. I want to Web-app based LLM for Law and Justice. My aim is provide these services to users:
Can you please help me in building it and dataset sources.
You might help people help you if you explicitly define in which country you desire to provide your services. Legal systems, customs, and practices are too diverse for a general solution.
In a previous lifetime I used an early commercialization of IBM Watson (no longer available) to build a demonstration of an intelligent assistant for pro se participants in US Courts. I always included warnings about the limits of use, even though that application found and returned relevant actual passages from relevant actual documents; it wasn’t generative. See
if you’re interested. Notice the caveats
“This is an informational handbook. This handbook is a guide for self-represented litigants. It is not legal advice and should not be considered as such. Do not cite to this handbook in your filings with the Court. The Court will not consider this handbook as legal authority.”
Generative LLMs are not reliable and foolproof sources of information and you should not represent them as such. Using any LLM to give legal advice seems like a quick way to be needing your own legal advice. Be very, very cautious. One of many examples where it is best if you hire a lawyer before you need one.
As far as sources, if you’re looking at the US market you should probably search the intraweb for sources like the ones mentioned here…
You should probably also look at what heavy players, such as Lexis, Westlaw, and Bloomberg Law are already doing with AI so you understand your competition.