Umar Butler

Background

About Umar Butler

I’m Umar Butler, a legal AI expert. I currently serve as the CEO of Isaacus, a legal AI research company building specialized AI models for the legal technology industry. As part of my research into legal technology and AI, I have published, inter alia, the world’s most popular semantic chunking algorithm (semchunk, downloaded over 900k times a month and used by Microsoft, IBM and The World Bank, to name a few), the largest open database of Australian law, the first semantic map of Australian law, and the first open LLM for Australian law. Previously, I was the Assistant Director of Data Science at the Attorney-General’s Department, the lead legal and justice policy department of the Australian government. I set up the AGD’s now thriving data science function and oversaw all national-level AI projects, from sentiment analysis of public submissions all the way to building a bespoke system comprised of 8 agents capable of classifying extremely long corporate statements with near-human accuracy in the span of 7 seconds. Before that, I was an AI engineer at Accenture, where I developed an AI legal assistant capable of answering highly complex queries related to Australian law with greater accuracy than the solutions of other major generative AI competitors (e.g, Claude 2, Bard, and ChatGPT), as well as an AI research assistant for an ASX 10 bank that could search for and download financial reports from the web and analyze them instantly. I hold a Bachelor of Laws (Honours) and Bachelor of Arts with a major in International Studies from La Trobe University. Through my studies, I was awarded a credit for having published a report with the Islamic Council of Victoria where I trained an LLM to identify anti-Muslim hate speech with 82% accuracy and used that model to find that there were at least 3.7 million Islamophobic tweets made between 2019 and 2021, and only 15% ended up being taken down a year later.