Industry Experts Share Their Thoughts on Generative AI
Three years used to be a good time frame for assessing the impact of technology on the various processes that drive legal and business outcomes. It’s not entirely clear if that’s a good measure anymore, but I thought it would be interesting to review the impact of generative artificial intelligence on e-discovery since it was hoisted upon us in 2022.
Healthy AI Skepticism
Some of my contemporaries have labeled me (unfairly, I think) as anti-generative AI. It’s entirely appropriate, in my view, to evaluate and characterize a person’s position or stance based on public pronouncements, writings, etc. And it is true that I’ve written and spoken about generative AI in recent years and it has not always been in glowing terms. Having worked in the legal profession for three decades, I think I have developed a sense of what works and what doesn’t, particularly when it comes to legal technology. To paraphrase my friend Doug Austin: I know few things because I’ve seen a few things.
But to be clear, I am not by any means anti-generative AI. I am, however, generally skeptical when new technologies claiming to solve the world’s e-discovery problems emerge. It is undeniable that the hype of generative AI as relates to e-discovery processes has been . . . what’s the right word? . . . fantastical? Until new technologies prove their value, I may appear to be a naysayer. It does not make me anti-generative AI. Frankly, in my experience, I believe this to be a healthy approach. I’m managing expectations.
I’m also not so full of myself that I’m not willing to admit I am wrong about something.
So, I thought it would be cool to revisit the past and sort of measure up where we are today, nearly three years since the public launch of generative AI.
Let’s Look Back . . .
About 18 months ago I wrote:
Because emerging technologies are by their very nature disruptive, it does not make sense to look only at the degree of disruption, the popularity, or even the ingenuity or innovative use of an emerging technology. Rather, it is essential to examine the specific existing and future workflows and use cases to fully understand the impact of a new technology.
To put this in context, I was writing about Generative AI and the Hype Cycle for a post on the ACEDS Blog that references the hype cycle that often accompanies new and disruptive technologies. I wrote at that time that I had not yet encountered meaningful, affordable, generative AI e-discovery solutions. And that’s not a criticism—I’m merely parroting what I’ve heard and seen based on user habits and experiences with the technology.
So, how if at all has my thinking on this evolved? There are applications that are beginning to show promise and move the needle, but I’m not aware of anything that has completely upended e-discovery processes.
And the truth is that I’m not in the weeds leading e-discovery projects anymore. I am privileged, however, to speak often with leaders in all segments of the e-discovery community, so I’m not entirely out of touch.
I Have Just One Question
For that reason, let’s not take my word alone as gospel on the impact of generative AI on e-discovery. I thought it would be interesting, though, to see what other industry leaders think.
What follows is a compilation of quotes from contacts across the e-discovery/legal industry who responded to me when I asked this question:
What has generative AI contributed to the e-discovery process?
Obviously, readers can form their own conclusions about the impact of generative AI on the e-discovery process. Following the quotes below, I offer my emerging thoughts.
I use Generative AI (GenAI) to summarize and extract insights from electronically stored information (ESI) that is otherwise difficult for humans to review and interpret efficiently. This includes formats like HTML files from social media accounts, which may contain usage data and other relevant information embedded in large volumes of unstructured text. By processing these documents through GenAI, I can quickly understand the key information they contain without manually reading through potentially thousands of pages. For example, I can ask targeted questions such as, “What are the date ranges of usage?” and receive accurate answers within seconds, eliminating the need for time-consuming manual compilation. Additionally, when working with long or complex documents, I can query the content directly and receive context-aware responses based on the document’s actual text. This significantly enhances the speed and effectiveness of document review and analysis.
Suzanne Clark
Mass Torts Discovery Counsel
Beasley Allen
The genie is out of the bottle, the AI toothpaste has been squeezed, and there’s no going back. Like the shift from typewriters to computers, or on-prem to cloud, generative AI is redefining and automating legal workflows. By delivering summaries, chronologies, privilege calls, and relevance ranking, it gives attorneys faster access to the facts and frees them to focus on strategy—not slog through data. Unlike TAR or CAL, GenAI and LLMs offer contextual understanding at speed and scale. And the e-discovery professionals staying current? They’re not just adapting—they’re leading the charge.
Brad Schaffel
Senior Manager
Sullivan & Cromwell
I could list all of the benefits of GenAI, but my simpler answer would be that this is a game-changing moment for e-discovery, and it feels a lot like the excitement we had when TAR first hit the scene … but this time, it’s even bigger. It’s exciting to be in an industry that’s constantly improving itself through technology.
Glen McFarlane
Managing Director, Cyber and Data Resilience
Kroll
Generative AI has added early value to e-discovery—accelerating document summarization, surfacing patterns, and drafting first-pass privilege or issue classifications. But in high-stakes legal contexts, this alone is proving to be insufficient. The next evolution—agentic AI—moves beyond static prompts to orchestrate tasks across identification, review, and production, while maintaining auditability and human oversight. We see generative AI as a foundational shift—but agentic AI is what allows e-discovery workflows to be truly scalable, defensible, and governed. The impact is not just speed—its structured intelligence applied where risk, relevance, and regulatory accountability intersect.
Kousik Chandrasekaran
Director of AI
Exterro
I have seen increased accuracy and substantial time and cost savings (estimated over $5 mil for one matter) in document classification, fact investigation and review when using generative AI tools. Simultaneously, I have appreciated how we have fostered a culture of genuine curiosity within the team and increased adoption of generative AI in various aspects of our role as advisors to the Firm and our clients. I’m also very thankful to my established community of legal technologists and friends who keep me focused on solutions that will have practical impact amidst a dynamic and evolving technology.
Stephen P. Dooley
Director of Electronic Discovery & Litigation Support
Sullivan & Cromwell
Generative AI’s contribution to the eDiscovery process is still in early days. To date, most of the use has been for review and analysis of data. The first set of capabilities focused on enabling users to “have a discussion with their data”. The second more complicated and ambitious use is to enhance review. Other EDRM stages are sure to follow. How GenAI will be used for those stages is hard to predict today – not because we lack imagination but because the underlying technologies are sure to continue to change dramatically in their scope and depth.
George Socha
SVP of Brand Awareness
Reveal
One of the most helpful contributions that GenAI has made to the ediscovery process is supplying document summaries that provide human reviewers with high-level overviews of long documents and lengthy email threads. The summary doesn’t necessarily help determine relevancy (although it can), but it does allow human reviewers to quickly determine if they can move on or need to spend more time reading through the document. Before GenAI, reviewers would have to skim through several pages of every document before determining how much more they would need to read. Not only do GenAI summaries cut down on the time of reading through documents in-depth, but the summaries can also help litigators gain a big-picture understanding of facts and issues involved in a case.
Brett Burney
eLaw Evangelist
Nextpoint
Generative AI is introducing exciting innovation to e-discovery by opening new ways to engage with complex data. Rather than relying solely on traditional search and review, teams can now experiment with AI-assisted summaries, privilege assessments, and pattern detection across large datasets. GenAI offers the potential to reimagine workflows, improve efficiency, and support more informed decision-making. While the technology is still evolving and being tested in practice, its early use signals a shift toward more dynamic, adaptive approaches in e-discovery, helping legal teams explore insights that were previously difficult or time-consuming to uncover.
Rose J. Hunter Jones
Partner
HILGERS GRABEN PLLC
Generative AI is giving legal teams a new kind of edge in e-discovery. It helps cut through the noise, highlight what matters, and bring clarity to complex data. That means attorneys can spend less time reviewing and more time thinking strategically. It’s not just about speed. Gen AI is about giving attorneys the confidence to make better calls, earlier. This creates space for the kind of work that clients remember.
Daniel Gold, Esq.
Principal, E-Discovery Managed Services Leader
BDO
Generative AI is rewriting the eDiscovery process, one point solution at a time. The low-hanging fruit was the first-level review, where generative AI immediately proved to be faster, less expensive, and almost universally more accurate than manual review teams or techniques such as TAR and CAL. This accuracy is a testament to the confidence we can have in the capabilities of generative AI. However, that is only the beginning of what will undoubtedly be a rewriting of the EDRM, as today’s new generative AI platforms have blurred the lines between processing, review, and analysis. Moving left, generative AI will eventually be embedded in corporate document storage and swallow identification, preservation, and collection. The current benefits are substantial, improving efficiency and creating a more level playing field in discovery processes. Yet, this development is poised to lead to even greater disruption, ultimately benefiting the end-clients the most.
Tom Palladino
President
eDiscovery.ai
Generative AI and agentic capabilities enable users to ask questions of their data in natural language, delivering powerful insights and outputs that were previously unattainable through traditional search or review methods. As one of the tools in your eDiscovery toolkit, GenAI facilitates the automatic classification of documents and the gathering of knowledge, allowing quicker access to information that can guide case strategy and can focus attention (and costs) on what matters most. This use of GenAI in eDiscovery – which highlights Review, Analysis, and Production workflows – is one of the most visible, exciting, and impactful additions to our field.
Tiana Van Dyk
Managing Director, Canada
Epiq
As clients look to reduce costs and improve quality of first pass review, GenAI has proven the importance of having a team of re-skilled review attorneys who are trained in this alternative workflow; attorneys who can support validation and QC at “right sized” costs to show value. Additionally, GenAI allowed us to create an agent most knowledgeable of all of our workflows, SOP’s and pointing to publicly available information on our specific tech stack, which has accelerated operational intelligence for our entire tech team and soon to be end users seeking quick tips on how best to use our tech.
Scott Milner
Partner and Global Practice Group Leader, eData Practice
Morgan, Lewis & Bockius LLP
Generative AI is shaking up e-discovery in the best way. It’s cutting through mountains of ESI by summarizing docs, spotting patterns, and surfacing what matters—fast. This use isn’t just about saving time; it’s about a more innovative strategy and reduced risk. And yes, it’s already driving change in ESI protocols, prompting conversations about how evidence is disclosed, validated, and defended in AI-assisted reviews. Essentially, Gen AI isn’t a new and shiny tool. It’s changing how discovery is conducted, from drafting protocols to managing productions. If you’re not paying attention, you’re already behind.
Maribel Rivera
VP of Strategy & Client Engagement
ACEDS
While some are using GenAI and dipping their toe in the water to incrementally accelerate document review and validation, the real promise is its ability to defensively call and review documents way upstream. When used to its full potential, the impact to timelines and legal spend are profound. This shift is starting to have an impact on how people think about the steps in the EDRM model, moving the tasks that take place from their traditional position to earlier in the workflow. I think that one of the barriers is a lack of understanding as to how the tools work or how to drive the tools. There is still a fear about the skills that are needed in order to utilize these tools. As systems become simpler, they will become more accessible to the lawyers and in-house eDiscovery teams that need to supervise them.
Rachi Messing
Co-Founder
Altorney
GenAI can streamline e-discovery across the EDRM lifecycle. During identification, GenAI tools can identify likely sources of relevant ESI within vast data populations. At preservation and collection, GenAI tools can help to identify and segregate potentially relevant documents, minimizing unnecessary processing. In review, this technology conducts rapid first-level responsiveness assessments, significantly reducing human workloads. GenAI also swiftly summarizes documents and mines key facts, enabling attorneys to focus on applying essential information to litigation issues. By automating time-consuming discovery tasks, GenAI dramatically improves efficiency, accelerates discovery timelines, and allows legal professionals to concentrate on higher-value case work.
Marcin Krieger
Counsel, Records & E-Discovery Practice Group
Reed Smith
Generative AI’s contribution to the e-discovery process has been most immediate in document review, including first pass review, QC, and privilege assessments, but its potential goes far beyond that. We’re starting to see generative AI play a role upstream in information governance, helping organizations manage ROT (redundant, obsolete, and trivial) data while identifying and monitoring SUN (sensitive, unstructured, and new) data in real time. Downstream, generative AI is emerging in arbitration and courtroom settings, from clause generation to drafting preliminary findings and streamlining submissions. Generative AI is poised to transform the entire EDRM lifecycle from end to end.
Monica Harris
Consult Harris LLC
As Manager of Discovery Services, I’ve seen Generative AI begin to influence how we approach eDiscovery—not by replacing legal expertise, but by enhancing how we apply it. With the right policies and oversight in place, we’ve found GenAI can make a meaningful difference, particularly in validation and reducing redundancy. Validation is where GenAI has shown promise. It helps us compare document sets and flag inconsistencies, which has improved the speed and reliability of our quality control—especially in areas like privilege review and responsiveness tagging, where accuracy is critical. We’ve also seen benefits in reducing redundancy. By helping prioritize more substantive content earlier in the review process, GenAI allows our teams to focus on what matters most. This not only reduces reviewer fatigue but ensures that the most relevant documents are seen first. That said, GenAI is not a standalone solution. It’s a tool that works best when paired with human judgment and clear boundaries, including a robust GenAI policy. When used thoughtfully, it can be a valuable partner in the eDiscovery process—supporting efficiency without compromising quality.
Stephanie Mills
Manager, Discovery Services
Cassels Brock & Blackwell LLP
A properly configured application of GenAI significantly improves the process of finding the documents that matter and doc review. With GenAI we no longer have to struggle to find the right key words or formulate Boolean search strings. Instead, we can approach a dataset in a more natural, conversational way. It’s now possible to analyze documents for responsiveness, privilege and issue tagging in a fraction of the time with accuracy that exceeds human review. GenAI outputs better inform counsel and review teams making discovery and doc review more effective and useful for achieving the client’s goals.
David Gaskey
Co-Founder & CEO
Altumatim
As can be seen above, everyone shares optimism about the prospects and benefits of using generative AI in e-discovery. And I’m sure this optimism extends to most e-discovery professionals. Unfortunately, I can only include here a limited number of responses from across the industry and I’m confident this does not encapsulate the views of everyone across the industry.
Survey Says!
But there are recent surveys that give some indication of the level of optimism. The 2025 State of the Industry Report produced by eDiscovery Today’s Doug Austin reported that 26% of the 551 respondents to the survey circulated in late 2024 expect generative AI to have a transformative effect on e-discovery. Nearly 82% of those respondents expect generative AI to create new workflows in 2025. The report also identifies the top use cases for generative AI, including document review automation, document summarization, privilege and PII identification, case strategy/ECA, translation, timeline creation, and deposition prep/summaries. Still, just over 12% said they were using generative AI in all or most of their cases, and 64% said they were using it in very few or none of their cases.
In March of 2025, the legal and business advisory firm Secretariat, together with ACEDS, produced their 2025 Artificial Intelligence Report. That report indicates that 60% of the mostly large law firm respondents to their survey are having “some issues” with generative AI. The top five use cases in this report included document drafting, internet search, legal research, document review and deposition/document summarization.
Lastly, Ari Kaplan, principal at Ari Kaplan Advisors LLC, who is also Chair of the ACEDS Advisory Board, an industry analyst and creator of the E-Discovery Unfiltered Report released earlier this year, interviewed corporate and law firm leaders on their use of generative AI in e-discovery. Over 60% of them report they are not using generative AI in connection with e-discovery. While the majority of those interviewed indicate strong optimism about the future of generative AI in e-discovery, the report also makes clear that there are strong reservations around privacy, security, accuracy and cost.
But what about the sentiments expressed by our industry leaders above? I’ve read through the responses to the question “Whathas generative AI contributed to the e-discovery process?” at least a dozen times now, and here’s what I think our experts are saying:
First, almost universally, our experts agree that generative AI accelerates time-consuming tasks like document review, summarization, and classification, which can reduce time spent on some legal tasks –not all of them e-discovery-related—and allow for increased focus on higher-value activities like case strategy.
Second, our experts frequently credit generative AI with reducing legal spend, sometimes dramatically, which is interesting given how little information is publicly available on the actual costs of using generative AI. But credit where credit is due – I can see how generative AI makes some legal tasks more cost-effective. But again, other than document review, a lot of these tasks are technically not part of the e-discovery process.
Next, there are several comparisons to past technological developments like technology-assisted review. I can see the parallels, for sure, but TAR and generative AI are very different technologies.
Finally, our experts acknowledge the future promise of generative AI technologies. They focus on how well it summarizes long documents, identifies relevant material faster, and uses natural language queries to give context and understanding to sometimes messy data. It strikes me, however, that these promises are focused almost exclusively on document review and not on other aspects of the e-discovery process.
In fairness, most experts agree that it’s early days for generative AI and the technology is still emerging. Like them, I am confident that an industry known for its innovation is going to hone and refine, test and validate until the technology offers sound, defensible tools that solve real-world e-discovery problems at cost litigants can afford.
Conclusion
So, in the end, a logical conclusion would seem to be that while generative AI is a promising new technology for use in e-discovery, the reality is that the technology does not yet appear to have had the sort of generational impact on e-discovery that one might expect.
One of the more interesting areas in which I personally see value for generative AI is in information governance and the left-side-of-the-EDRM. A few of our experts agree that this is an area ripe for development. I can see a world in which agentic processes more effectively corral an organization’s data early in the e-discovery process, helping to get an understanding of the relevant information early and thereby reducing the volume of data in the first instance, thus streamlining the entire e-discovery process. I’m entitled to dream, right?
Three years ago, no one had given much thought to generative AI, least of all whether it would impact e-discovery processes. It hasn’t, in my view, changed the process very much. That’s not me being a naysayer, I’m merely reporting the facts. There have been contributions that have made tasks –mostly in the document review area—more efficient, but wholesale change I do not see. That said, there’s promise and lots of optimism on the horizon.
I, too, am optimistic. I’m not sure we’ll ever achieve “easy button” status in e-discovery because there’s always going to be a need for a human in the loop when it comes to matters involved in the legal system, but I look forward to the day when the drudgery and pain points of large data collections are more easily managed.
[View source.]
