'I’m Having Flashbacks': The Evolution of Legal Document Review From Paper to Generative AI

Let’s take a stroll down memory lane and look back at how far we’ve come, not just for nostalgia’s sake, but to ease the anxiety, uncertainty, and skepticism many of us are feeling right now. Take a deep breath, my friends, as we have been here before. And just like then, we will figure it out together. 

I just returned from my 20th Legal Week or LegalTech if you have been around long enough to remember the original name. One thing was crystal clear: the theme dominating every session and every corner of the Exhibit Hall was AI, AI, and more AI. As I wandered between sessions, I found myself having flashbacks, moments from years past when we, as legal professionals, had to lean in and adapt to new waves of technology. 

Let’s take a stroll down memory lane and look back at how far we’ve come, not just for nostalgia’s sake, but to ease the anxiety, uncertainty, and skepticism many of us are feeling right now. Take a deep breath, my friends, as we have been here before. And just like then, we will figure it out together. 

The 1990s—The Paper Era 

I started my legal career in the early 1990s. Legal document review was a strictly manual process, requiring physical review of printed documents in large-scale litigation and regulatory investigations. I don’t know about you, but my colleagues and I spent many days in dusty warehouses wearing gloves for weeks or months, reviewing thousands, sometimes millions of pages. 

I am sure some of you can remember when we had multicolored Post-it notes, and each color represented a witness. Reviewers manually flagged relevant documents, often leading to inconsistencies, and important evidence was often missed due to fatigue, oversight, or lack of standardized processes. Some documents had 10 different colored Post-its, and you had to pray that those Post-its didn’t fall off before the copy center made the copies for the witness files. Oh, and what about bates stamping? Yep, can you imagine: click, click, click every page by hand. This was my life for years prior to the technological insurgence in the late ‘90s. 

Even with all this manual work, this foundation and appreciation had many benefits that set the stage for what was to come. Teams worked in war rooms, physically sorting and categorizing documents and discussing hot docs and key issues they found, which ultimately became the early model for collaborative legal review. Law firms refined best practices for document review, setting the stage for structured workflows. Paralegals played a crucial role in developing coding sheets and organizational methodologies to classify documents. 

The 2000s—The E-Discovery Boom: A Digital Shift With Growing Pains 

Someone heard our prayers with the invention of scanning to electronic format and the paralegal savior of electronic bates stamping. I know some of you are smiling right now thinking of it! Remember the terms “blowback,” “logical/physical unitization,” and “OCR”? The introduction of early document management and review tools like Advanced Revelations, Summation, and Concordance were revolutionary. 

Legal teams started to embrace these tools, and the litigation support team was born. My first career change happened as I transitioned from a paralegal to a litigation support manager and my new career was born. I remember having the same conversations with attorneys as I have now. I would suggest to them, “lean in and embrace the technology,” yet they were hesitant. Everyone’s nerves set in as to what the future of this technology would mean for our careers (Doesn’t that sound familiar?). 

In the early 2000s, “e-discovery” entered the scene. The transition from paper to digital documents accelerated during this time due to the rise of electronic records. The digital shift led to an overwhelming surge of emails, PDFs, spreadsheets, and corporate databases that contained massive amounts of unstructured data. Early software solutions (e.g., Concordance and Summation) relied on basic keyword searches, which often returned too many irrelevant results or missed key documents. Legal teams could now electronically search and tag documents, reducing review time. Law firms began using metadata analysis to refine searches based on date ranges, email senders, and document types. This marked the first step toward automation in document review. This transition to scanning, storing, and indexing digital documents required law firms and corporate law departments to invest in new software, training and IT support. 

As a litigation support professional, I was the rockstar. Every attorney came to me and my team for advice on how to navigate this unknown. It was an exciting time for me, preparing programs to educate lawyers and paralegals on these arising topics. How do I review these emails? What do I have to preserve? How much does it cost to restore backup tapes? I can remember it was $2,500 per gigabyte to process data, and only three companies did it (you know who you are). I remember those discussions: “Oh, this isn’t a big case, we don’t care about metadata!,” or “I’m not a technical person, I’m a lawyer.” Wow, how things have changed. The point is that we were all panicking, and guess what, we leaned in and survived. 

The e-discovery industry truly gained traction in 2003 with the landmark rulings in Zubulake v. UBS Warburg. This case produced five separate decisions, three of which had particularly significant implications for the handling of electronic evidence. These decisions clarified the duty to preserve electronically stored information (ESI), distinguished between accessible and inaccessible data, and established a framework for cost-shifting in e-discovery. Collectively, they laid the legal foundation for modern electronic discovery practices and highlighted the growing importance of digital information in litigation. 

The 2010s—The Rise of AI and Predictive Coding: A Game Changer 

By the 2010s, technology-assisted review (TAR) and predictive coding transformed document review by applying machine learning to categorize and prioritize documents. Courts and legal professionals were hesitant to trust AI-driven document review and early legal battles questioned whether TAR was legally defensible compared to manual review. Attorney resistance came from their fear that machine learning models were “black boxes” that they could not fully understand or control, and AI models required extensive human input to teach the system what constituted relevant or privileged material. 

However, machine learning did increase review speed and accuracy. TAR automatically categorized documents, significantly cutting review costs and time while AI-assisted analytics grouped similar documents, allowing legal teams to focus on relevant evidence faster. 

As TAR gained traction, the resistance from some legal practitioners, courts, and opposing parties, led to legal disputes and court rulings that shaped its acceptance. Judge Andrew Peck issued the first U.S. opinion endorsing TAR, calling it an “acceptable way to search for relevant ESI in appropriate cases.” This was groundbreaking. More cases followed, reinforcing AI’s role in e-discovery, but the “TAR wars” continued. 

While TAR wars continued throughout the decade, a new era of document review also began with the introduction of cloud-based e-discovery review platforms like Relativity, Everlaw, and DISCO, enabling teams to work from anywhere. Security, breaches, and attorney-client privilege protection were the leading topics of concern, and, yet again, these issues were faced with hesitation, anxiety and skepticism. 

Today—The AI Takeover 

 
So here we are, and the 2020s have ushered in the age of AI-driven automation, including generative AI, large language models (LLMs), and intelligent automation. Generative AI holds significant promise for streamlining legal document review as it can summarize, classify, and analyze vast quantities of data at speeds far beyond human capacity. Unlike traditional TAR, Gen AI models like GPT and others go beyond keyword matching and classification by generating summaries, identifying key legal issues, flagging potential privilege or sensitive content, and even proposing redactions. 

For law firms, corporate legal departments, and government agencies, this means faster turnaround times, lower costs, and potentially higher-quality results. Gen AI is especially valuable in early case assessment, second requests, internal investigations and compliance reviews. 

The judges panel at Legal Week encouraged the responsible use of technology to reduce the burden of discovery, emphasizing that while AI can assist, it cannot replace legal judgment. In recent months, we have seen court rulings reminding lawyers that they are responsible for any filings or document productions generated or assisted by AI. Some judges have already sanctioned attorneys for submitting AI-generated content without proper verification. 

Given this guidance, will there ever be full-scale adoption of AI in the legal profession? If we follow the trends of the evolution of technology-assisted document review, my prediction is it is not a question of if, but when. Realistically, we are in the early majority stage. Some forward-thinking firms and legal service providers are already integrating Gen AI into their review platforms with high success rates, while others are watching closely before making the leap. Over the next few years, we will likely see widespread adoption, especially as tools become more reliable, auditable and integrated into existing review platforms. The timeline depends on more case studies, judicial precedent, and clearer guidance from the bar. 

I hope you enjoyed this trip down memory lane. Despite the challenges, technology has undeniably transformed the legal review process, making it faster, more efficient, and far more cost-effective—a dramatic evolution from the paper-based workflows of the 1990s. 

I’ll leave you with the words of my teenage heartthrob, Rob Lowe, who summed it up perfectly during his keynote at Legal Week: 

“AI makes Eli Whitney’s cotton gin look like nothing,” he said. “I mean, it’s the Industrial Revolution on steroids. I mean, it’s the internet times everything. We all know it. You should not fight it because the genie is out of the bottle.” 

Christa Iannone is a trailblazer in the litigation support space, guiding law firms and corporate law departments at the intersection of law and technology for 35-plus years. Iannone is the senior technology adviser at The MSC Group and can be reached at 2152825669 or Christa.Iannone@themcsgroup.com. 

Reprinted with permission from the May 2nd edition of the “Legal Intelligencer" © 2025 ALM Global Properties, LLC. All rights reserved. Further duplication without permission is prohibited, contact 877-256-2472 or asset-and-logo-licensing@alm.com.