ChatGPT Can Already Enhance Your eDiscovery Process – With the Right Expertise
We look at how ChatGPT can enhance your eDiscovery process.
Aaron Crews, Chief Product and Innovation Officer, UnitedLex
You’ve undoubtedly heard and read a lot about OpenAI’s ChatGPT – probably both good and bad. ChatGPT is so good at writing essays that several schools and colleges have banned its use. The latest version of ChatGPT (GPT-4) even passed the Bar Exam – not just barely, but by “a significant margin.”
People have already turned to ChatGPT for a variety of use cases including creating articles*, drafting emails, formulating product descriptions, summarizing articles and publications, even writing poems, coding computer programs and composing music! ChatGPT even served as the officiant in a recent wedding!
However, there have also been significant issues with ChatGPT as well. There have been data privacy concerns about the product and several lawsuits filed against OpenAI for libel and copyright infringement. Not to mention, the FTC recently announced it was investigating OpenAI for potential consumer protections violations.
Lawyers have also encountered issues trying to use ChatGPT in their work. A court in the Southern District of New York recently imposed Rule 11 sanctions including a $5,000 penalty against two lawyers that submitted several case law citations as research for a brief that were determined “to be bogus judicial decisions with bogus quotes”.
As a result, several Federal court judges (including in Texas, Illinois and Pennsylvania) have issued standing orders requiring mandatory certifications by attorneys on whether they used AI and, if so, what they did to verify the results. These judges are expecting lawyers who use ChatGPT or other generative AI technology to demonstrate the expertise to use it properly and train and test the results.
Enhancing eDiscovery with ChatGPT
The good news is that ChatGPT has already emerged as a powerful tool to streamline workflows, improve efficiency, and tackle common challenges faced by legal professionals. Legal technology companies, such as UnitedLex, are already adopting this innovative method to enhance their eDiscovery processes. Here are some of the ways in which ChatGPT can revolutionize your eDiscovery experience:
Managing Data Volume
Let’s start at the beginning of the EDRM life cycle. With the explosion of digital data today, managing and sorting through all this information can be a challenge, especially with as much as 85% of data in organizations being either Redundant, Obsolete, or Trivial (ROT). ChatGPT can help to manage these large volumes of data to help remediate ROT data as well as extract relevant data more efficiently.
Early Case Assessment
ChatGPT can analyze and summarize the content of a legal complaint, including identifying and listing dates, names, and other concrete facts mentioned in the complaint, as well as identify and flagging keywords or phrases that are relevant to the case. This analysis can facilitate early case assessment to make strategic decisions that maximize the potential for success.
You can even apply ChatGPT’s analytical capabilities to compare the current case to others and gain insights on similarities and differences. This type of comparison enables your legal team to strengthen arguments and refine case strategies.
Identifying Patterns and Anomalies
ChatGPT can quickly scan and analyze data collections to identify patterns, trends, or anomalies that might be relevant to a case. For example, it can identify unusual email communications or financial transactions that could suggest fraudulent activity, which can streamline the investigation process.
As I discussed during a recent webinar presentation, with these generative texts, it’s essentially two algorithms playing together (a ping-pong game, if you will) where one player is building and the other player is either validating or rejecting the other player’s suggestions. The process runs in rapid succession until an agreement is reached.
Craft Effective Keyword Searches
ChatGPT can expedite the process of identifying relevant keywords and constructing effective Boolean searches. This process includes engaging in a conversation with ChatGPT to receive suggestions for keywords, refining search queries, and generating comprehensive Boolean search strings. This iterative process can lead to enhanced keyword searches that efficiently uncover key documents in the case.
Streamline Document Review
Document review is still the most time-consuming and expensive phase of eDiscovery. ChatGPT can help to streamline document review in two ways: 1) It could generate guidelines or criteria for what types of documents might be responsive to each request in the request for production, and 2) It can perform an initial review of documents and identify relevant documents based on criteria, such as keyword search, semantic content analysis, etc.
Easier Said Than Done Without Expertise
As Chief Product and Innovation Officer at UnitedLex, I understand how ChatGPT can help support the above eDiscovery use cases and we’re continuing to learn more about its capabilities and limitations to support eDiscovery every day. In fact, if you ask ChatGPT how it can help streamline eDiscovery, it will identify many of these same eDiscovery use cases that I have outlined here.
Of course, ChatGPT will also tell you it can help with “comprehensive legal research” and “finding relevant court cases and legal opinions”. We all saw how that turned out in the New York sanctions case mentioned above.
And that’s the point. Using ChatGPT to enhance and streamline the eDiscovery process isn’t a “Staples EASY button”. It involves training and testing the model using machine learning techniques, followed by fine-tuning to further optimize ChatGPT’s performance for eDiscovery tasks.
The actual technical implementation of passing the data to ChatGPT for training, testing and fine-tuning can vary depending on the platform or tools being used. Common approaches include:
- Training Pipelines: These pipelines typically involve scripts or commands to process and feed the annotated data to the model for training and fine-tuning.
- APIs and SDKs: Some platforms or tools offer Application Programming Interfaces (APIs) or Software Development Kits (SDKs) that enable developers to integrate ChatGPT and perform training and fine-tuning programmatically. The APIs allow data to be sent in a specified format to the model for training and fine-tuning, and the responses or updates can be retrieved for further processing.
- Custom Scripting: In certain cases, custom scripting or coding may be required to interact with the ChatGPT model and perform training and fine-tuning. This can involve writing scripts to preprocess the data, make API queries to the model, and conduct the training and fine-tuning process.
Any AI algorithm – including ChatGPT – without effective training, testing and fine-tuning will yield flawed results – at best. ChatGPT is still just a tool – one that requires expertise to be used properly to achieve the desired result. It takes experienced people executing well-designed processes to fully leverage ChatGPT – just like any other technology.
Conclusion
Lastly, it’s important for eDiscovery practitioners to be cognitively flexible – having the growth mindset to think about how to deploy these tools to solve problems you run into all the time. For example, at UnitedLex we are able to look at corporate legal data and leverage these technologies to help clients. In litigation, the tools look at the core corpus of your evidence, and suggest case themes you can present to a jury, for example, to make a very persuasive argument about why you should win. Smart practitioners will grab these opportunities and make use of them.
ChatGPT is an amazing technological achievement and by leveraging ChatGPT’s capabilities effectively, legal professionals can accomplish amazing things. Doing so is easier said than done, but with people that have the right expertise, sound processes and ChatGPT’s terrific technology, your team can overcome common eDiscovery challenges and achieve more efficient and accurate results.
*Not this one though. It was all me, except where indicated!
For more information on UnitedLex’s Litigation & Investigation Services for Corporations, click here.
Resources:
These Schools and Colleges Have Banned Chat GPT and Similar AI Tools
https://www.abajournal.com/web/article/latest-version-of-chatgpt-aces-the-bar-exam-with-score-in-90th-percentile
https://people.com/chatgpt-officiates-colorado-wedding-7555337
Suit Filed Against ChatGPT for Libel Due to Hallucination: Artificial Intelligence Trends
OpenAI is Now Getting Hit With Copyright Lawsuits. No Joke!: Artificial Intelligence Trends
https://www.documentcloud.org/documents/23856993-sanctions-mata-v-avianca?responsive=1&title=1
https://www.veritas.com/news-releases/2016-03-15-veritas-global-databerg-report-finds-85-percent-of-stored-data
