Answering the “Big Three” AI Questions for Lawyers: AI 101
August 2023
By
Chris Frickland
In Case You Missed It: Recapping our latest LLM-informed webinar and providing a primer for all lawyers on the basics of AI.
In partnership with LegalTech expert Zach Abramowitz, Axiom has been exploring AI through Large Language Models (LLMs) – an exploration that began following the launch of ChatGPT. We understood immediately this technology would become a very significant issue (with both positive and negative implications) for legal departments. We also understand that AI-informed technologies can have an immediate impact on the daily work habits of lawyers – if they can harness the tools effectively.
If you’re a tech novice, this primer is for you as we ask and answer the Big Three AI questions:
- What is Chat GPT? How does this differ from previous AI technologies and why is it important for my profession?
- What’s the difference between AI, LLMs, and ChatGPT? How can I equip myself with the knowledge to comfortably converse about LLMs in my legal department or law firm?
- How can I apply practical advice on how to use ChatGPT and other LLMs in my day-to-day work?
Question 1: What is ChatGPT, and why does this matter to me?
Almost everyone has heard about AI and you’ve also probably heard of ChatGPT by now. But where it gets a little confusing is the terminology around large language models. Maybe you’re a little embarrassed to ask how exactly ChatGPT differs from previous technologies or LLMs. Here’s what you need to know:
- ChatGPT is a natural language processing tool driven by AI technology. In more laymen’s terms, it allows you to have human-like conversations (and much more!) with the chatbot. The language model can answer questions and assist you with tasks, such as composing briefs and emails.
- ChatGPT is the fastest-growing application to ever hit 100 million users. Let that sink in. To put it in perspective: Instagram took two and a half years and TikTok took nine months to reach that metric. ChatGPT took less than one month.
- Your colleagues are using ChatGPT. This is especially true if you’re in the professional services industry. Why is this so? Up until now, there's often been an assumption from many professionals in white-collar jobs that AI would affect manual labor before it impacted “intellectual labor.” The general thought was that knowledge workers were going to be displaced only after roles like truck or taxicab drivers and factory workers. But in reality, it turns out that what's made ChatGPT so unique is that if there is a supposed labor “threat,” the knowledge workers are now worried, which includes academics, lawyers, accountants, consultants, etc.
Question 2: What is the Difference Between ChatGPT and a Large Language Model?
Simply said, ChatGPT is a Large Language Model (LLM) that uses AI.
OpenAI essentially created LLMs including ChatGPT. Now, there are many different LLMs (like Google’s Bard), and there will certainly be more in the future from other major technology players (Facebook, for one). Essentially, the architecture of an LLM is that you feed a huge amount of data into it to learn how to predict language.
Important notes about LLMs:
- The bigger the parameters, the better the accuracy/output. ChatGPT3.5 had 175 billion parameters. It’s rumored that ChatGPT4 will have one trillion parameters – meaning the technology is getting more and more accurate and more and more useful as a tool to support daily work.
- LLMs are expensive to operate, which can slow down the availability of models. OpenAI turned to Microsoft since their products are costing them millions of dollars a day, which will get better over time but for now is a significant factor.
- Reinforcement Learning Human Feedback (RLHF) was the game changer for LLM’s speed to launch. This is the key that got the more primitive GPT3 over the hump to become what we know now as ChatGPT. When the company introduced actual human beings to cross-check AI and correct inaccuracies, their AI platform took off.
Want to sound more knowledgeable around the water cooler? Remember this as the theory: “Language goes in, language goes out.” Now you understand all the technological lingo, but the question remains: how can you adopt this technology to aid your own career on a day-to-day basis?
Question 3: How can I leverage AI in practice?
From a very early stage, after the release of ChatGPT, the legal industry became identified as one of the primary targets for disruption. And what's interesting to note is the founder of OpenAI, Sam Altman, said one of the first tasks in which an LLM must become proficient for specific tasks is legal. And Goldman Sachs says it will be; in a recent report, the firm forecasted that generative AI tools like ChatGPT will accelerate legal industry automation so much so that 44% of all current manual legal tasks may soon be automated.
Lawyers must start considering how LLMs might impact their daily work. The easiest way to do so is to make LLMs like ChatGPT your default tool for personal and administrative tasks. Unless you’re a pro at using this already, and especially if your company has an explicitly stated policy on how to use it, refrain from the risks involved in using these tools for core legal work. Instead, use it to make your life simpler and more efficient. Will this replace a Harvard-trained lawyer? Of course not. But it will make your most frustrating tasks just a bit easier to tackle.
Here are four ways lawyers can use AI in practice:
- Use it for routine, time-consuming administrative functions. For example, copying text from a PDF into a Word doc can be incredibly frustrating. ChatGPT can clear up that formatting without you manually making the anticipated edits. Or if you need content from a Word document to be transferred into a table, you can use AI there as well.
- Ask it to summarize client emails for you. If you’re overwhelmed with your inbox full of long emails with various requests, you might miss out on something important if you don’t spend the time fully reading and absorbing the content in them. Instead, consider asking ChatGPT to read your longer emails to find what the major request is from the sender. You’ll be more efficient and more conscientious in your email replies.
- Improve your to-do list by prioritizing appropriately. If you provide AI tools with your personal and professional to-do items or upcoming meetings and project requests, you can ask it to sort by priority and have your calendar now optimized for the most efficient use of your time. You’ll end up each day with an actionable item list that is best suited to your specific needs.
- Get the best results by becoming an expert prompt engineer. Basically, this means providing as much information as possible to the LLM, thereby increasing the chances to get a more refined result. How do you build the perfect framework for prompt engineering? Through these six essential steps:
-
- Context: Set the stage with relevant background and perspective
- Task: Clearly define the specific assignment
- Purpose: State the ultimate goal or objective
- Criteria: Define measures of success
- Resources: List available or needed materials/tools
- Examples: Provide models for inspiration and clarity
-
An example you can use:
“ChatGPT, draft me a contract. Here is the information needed. Here is what my purpose for this contract should be and what I want you to do. Here’s why I need this contract. Here are some examples of previous contracts I’ve done.”
Of course, AI does not replace your legal skills, but its help with your related administrative tasks is limitless.
💡 Learn how to better manage the unique challenges and complexities presented when onboarding, utilizing, or developing artificial intelligence.
Posted by
Chris Frickland
Chris heads the Technical Program Management and Data Science functions at Axiom. As an inaugural member of the Axiom Research and Development Team in 2018, Chris has delivered all Axiom’s platform initiatives, spanning internal tools for matching Axiom’s global legal talent to its prestigious clients, a digital experience for Axiom legal talent to manage their end-to-end experience with Axiom, and Axiom’s first push to present its black book of legal talent on the web to current and prospective clients. In addition, Chris spearheaded Axiom’s introduction of machine learning technology, pioneering the use of Axiom’s vast data sets built from years of manual talent-to-client matching to recommend new and unknown talent to Axiom’s clients. The resulting “Magic Lawyer Finder" technology aided Axiom’s record 2021 growth in an extraordinarily talent supply-constrained environment. Prior to joining Axiom, Chris held technical leadership positions at Tableau, Starbucks, and BBI Engineering, enabling him to dive deep into technical challenges in museum-quality technology installations, city-scale infrastructure, and enterprise-scale data analysis.
Related Content
AI in Law: Six Ways Axiom is Riding the AI/LLM Wave (and how you can, too)
AI in law is changing the legal landscape as companies & law firms grapple with using AI. Here's how Axiom uses artificial intelligence in practice.
2023 in Review: Top 13 Most Popular Resources for In-House Legal Teams
Check out Axiom's comprehensive guide to the most popular insights over the last year, including transformative trends shaping the legal landscape in 2024.
AI Revolution for In-House Legal Teams: The Seven Headlines All Lawyers Need to Know
Learn more as we delve into the current state of the AI revolution and its profound impact on legal departments in 2023.