With the release of GPT4, and in particular with the demo video showing ChatGPT4 being used to answer a question about tax laws, people have been asking my thoughts about whether Rules as Code is still necessary. Should we be getting human beings to deal with encodings of legislaton, or should we just let large language models do the legal reasoning? Are there parts of the Rules as Code process that should be dealt with by things like GPT4?

LITCon 2023 On Monday I had the opportunity to participate in LITCon 2023, a conference put on by the Legal Innovation and Technology Lab (LITLab) at Suffolk Law. LITLab is run by David Colarusso, who was the person who first introduced me to declarative logic programming for legal knowledge representation years ago. Perhaps my favourite part of the conference was the presentations done by LITLab students on the projects that they had built.

Wow, it’s been nearly a year since I updated the blog? Wild. First thing to note, the address has changed. Now that I understand better how to use multiple email addresses with the same GitHub account, I’ve moved this blog to my personal GitHub account, so you will now find it at https://gauntlet173.github.io/. Assuming Microsoft doesn’t become some sort of evil overlord, it should be able to live here for a while.

This post describes a method that I have implemented in v1.3.15-alpha of Blawx for using abductive reasoning features of the s(CASP) programming langauge to power relevance decisions in expert systems. I’ll try to give you a brief introduction to what problem we’re trying to solve, how it has been solved before, how I have solved it in Blawx, and how those solutions differ. The Problem An expert system is a sort of application that takes a knowledge representation, in most cases a declarative logic description of a set of rules, and a question that it has been asked to answer, and then attempts to answer the question.

Long time no blog! There have been lots of big, exciting changes to Blawx over the last while. I’m now working on getting the documentation caught up. When that’s done, I promise I’ll come back and do a big post about Blawx. Today, I just want to talk about an idea I’ve been developing over the last few days. What’s Phase 2? When we talk about Rules as Code, we tend to talk in abstract terms.

It’s been a while since I did a blog about the changes to Blawx, so let’s get you caught up! Spans For reasons that will be clear shortly, we needed a way to delineate and name parts of a paragraph of text. So I made some changes to CLEAN so that you can use the following format to be able to treat spans of legal text as though they were a sub-part of the law.

Rather than me repeating it here, you can check out my live tweet of the final session. My Thoughts I’m not sure if I’m surprised, or not, but there was almost no consensus on anything, except that we should experiment concretely, not debate abstractly, the tools should be explainable, and the products of the work should be publicly accessible and usable, not behind proprietary technologies or pay walls. Honestly, at LEAST getting consensus on the need for explainability is a step in the right direction.

TL;DR There is a new version of Blawx, live online, that allows you to edit legislation in a markdown-like language called Clean. I will be participating in a conference on Monday on Rules as Code 2.0, looking at how we generate the skilled workforce that rules as code requires. Registration is free and highly recommended. Videos from the SMU Computational Legal Studies conference are available online. Blawx Update I’m working with the Canada School of Public Service on developing Blawx as a platform for learning projects in the Rules as Code space.

The Problem I made a number of changes to Blawx that makes it possible to organize your code according to the pieces of the legislation. This is the first of a number of improvements that are all designed to let you use Blawx to do a structurally-isomorphic encoding of the legislation. The problem was that the initial version did not provide any good way to edit the Akoma Ntoso source code for the legislation.

There was a conversation on Twitter this week that helped to clarify my thinking about something, and I wanted to share. There is a sort of tension between “code” and “no-code” (sometimes “low-code”) solutions. People who consider themselves software developers see people trying to use no-code solutions to do things that the no-code solutions are not good at, and they get frustrated that the capabilities of those tools has been overestimated.