Friday, December 6, 2024
spot_img

The Evolution of LegalAI and Canada’s Alexi

With Goldman Sachs predicting that 44% of legal jobs will be impacted by AI, the race for the dominant solution in this space is fierce. Legacy players like Lexis Nexis and Thomson Reuters have struggled with boil the ocean approaches that fell victim to generative AI’s hallucination problem. But an up and coming Canadian contender with thousands of users has taken a different approach, begun before the launch of ChatGPT and not generativeAI dependent, and is changing the conversation and the landscape significantly. Alexi, a Toronto-based legal tech AI startup and other similar legal AI tech, may even make legal services more accessible to those for whom it is currently cost prohibitive. 

Alexi, which has between eight and a dozen full time lawyers as part of its development team (“they work in product development full time and are not billable,” said founder and CEO Mark Doble) began using AI to develop research briefs, and has since progressed to mapping out every stage of the litigation process and identifying the most significant case law attached to each stage. With over 20 million question and answer pairs for case law in Canada and the US, “our competitive advantage and sole focus is retrieval,” said Doble. “We use some generative AI after the research phase is complete, but our focus on retrieval is what makes us different and results in not having to deal with hallucinations. Zero.” 

RAG before RAG was RAG 

The legal profession, long viewed as resistant to technological change, has seen a rapid transformation thanks to the rise of AI tools specifically designed for the field. In the past, junior associates or legal clerks would spend hours or even days compiling relevant cases and statutes to support legal arguments. However, with the advent of Alexi, lawyers have been able to produce comprehensive legal research in a fraction of the time.

This move toward AI-assisted legal work stemmed from the need to streamline the mundane yet time-consuming aspects of legal research. Alexi, early on, identified this gap and provided a solution. But this was only the first step in what would become a more expansive toolset for legal professionals. Alexi’s proprietary approach to information, essentially deploying RAG before it became widespread, is a critical differentiator. 

“Generative AI is Not the Focus”

RAG, or Retrieval-Augmented Generation, is a framework in AI that combines retrieval techniques with generative AI models to improve the quality, accuracy, and relevance of generated responses. In simple terms, RAG enhances generative models by allowing them to “look up” or retrieve relevant pieces of information before generating a response. This approach is particularly useful in scenarios where up-to-date, specific, or factual knowledge is needed.

Limiting the data used with this approach has allowed Alexi to evolve beyond simple research assistance and become an essential resource for legal professionals seeking information on precedent. “We don’t think we’re going to need to expand beyond this information set except as new decisions come up,” added Doble.

This knowledge set has been the key in ensuring that Alexi’s recommendations are both comprehensive and reliable. With each query, Alexi becomes smarter, allowing it to provide more nuanced and tailored responses. This volume of data has transformed Alexi into not just an AI tool, but a vast legal database with near-instantaneous access to legal knowledge that would take human researchers weeks to sift through.

From Research to Pleadings

Alexi has expanded its capabilities into more sophisticated areas of legal work, and is focused next on a much more complex area of the law, the drafting of pleadings. Pleadings, which are formal written statements submitted to a court, are the backbone of litigation. Alexi’s evolution into assisting with pleadings shows its trajectory from a tool for preliminary research to one that supports critical stages of litigation. Competitors like Lexis Nexis and Thomson Reuters have gone with breadth and generative applications, versus Alexi’s specificity and focus on retrieval. LexisNexis data comes from more than 50,000 constantly updated sources, such as court filings, law firms, news sources, and websites. 

Alexi’s approach of limiting information retrieval to a highly contained set of case law data, not driven by generative AI but the company’s proprietary dataset and approach to retrieval, is proving highly effective. Thousands of legal professionals are now on the Alexi platform. The platform’s ability to support both the research and drafting processes highlights its commitment to optimizing the entire legal workflow. By moving from research briefs to pleadings, Alexi’s evolution mirrors the changing needs of the legal industry, pushing the boundaries of what AI can do in legal practice.

How far is Alexi ahead of its competitors? In August, Lexis Nexis relaunched its offering after studies showed its dependence on generative AI and LLMs, including Anthropic’s Claude AI assistant and GPT-4 on Microsoft Azure, were resulting in an unacceptable percentage of hallucinations. “We have no hallucinations,” said CEO Doble. “Our architecture doesn’t allow for it.” Lexis Nexis has now shifted to a retrieval augmented generative engine to eliminate hallucinations, but hallucinations persist, according to a Stanford study. “While both hallucinate less than a general-purpose AI tool such as GPT-4, they nevertheless each hallucinate more than 17% of the time, the study concluded. The study also found substantial differences between the LexisNexis (LN) and Thomson Reuters (TR) systems in their responsiveness and accuracy, with the LN product delivering accurate responses on 65% of queries, while the TR product responded accurately just 18% of the time.”

What is the potential impact of AI on how the law is practiced? Will AI result in lower employment in the sector? Doble doubts it. “There’s an enormous unmet need for legal services, and the preventative factor is most often cost. We believe this technology will make legal services accessible to people who need and can’t afford it. Only about 15% of possible lawsuits actually proceed.” Given that access to legal services has historically been limited to the wealthy, this could be a boon to many who feel financially oppressed by the legal system. “We feel this could change the dynamics of the industry and open up access,” added Doble. 

For thousands who have found themselves without representation as legal aid funding has changed in jurisdictions like Ontario, AI could soon put the legally disenfranchised on more equal footing. As the tech moves into automation of pleadings, a much more complex application, the implications for the industry will become clear. 

Featured

Combating Counterfeits: Open Commerce Platforms Redefine Brand Integrity in Digital Marketplaces 

By Justin Floyd, Founder and CEO, RedCloud Technologies In an increasingly...

Building a Business on Your Own Terms

Fatima Zaidi is the CEO and Founder of Quill...

Maximizing Business Efficiency: The Role of IT Consultancy in Glasgow

In today’s rapidly evolving business landscape, technology plays an...

How Charities Can Manage Enormous Public Money Dumps

Pexels - CC0 License Charities and nonprofits are critical for...

5 Experts To Help You Navigate Divorce

Image credit No one wants to think that their marriage...
Jennifer Evans
Jennifer Evanshttp://www.b2bnn.com
principal, @patternpulseai. author, THE CEO GUIDE TO INDUSTRY AI. former chair @technationCA, founder @b2bnewsnetwork #basicincome activist. Machine learning since 2009.