Expectation vs reality: what generative AI really offers eDiscovery

One of the biggest questions legal and investigative firms are currently pondering is whether or not Generative AI really is the next frontier in eDiscovery. It’s certainly making enough waves in other business sectors. But how much of the GenAI hype is grounded in reality? Are the results really worth embracing yet another big change?

Hype cycles and the road to productivity

As Gartner so clearly illustrates in their Hype Cycle diagram, new technology almost always hits the market with a flurry of excitement and inflated expectations. This is typically short-lived, giving way to a so-called “trough of disillusionment” as early adopters lose momentum and the mainstream market remains on the fence.

The challenge for technology vendors is to bring practical, mainstream applications of the technology to the market in order to bridge the gap between initial expectations and the technology’s genuine potential, without falling into the “trough of disillusionment”. (Read more on this phenomenon in Geoffrey Moore’s “Crossing the Chasm”.)

The peaks and troughs of AI’s initial hype cycle have been well documented, with practical applications now enjoying widespread mainstream appeal. GenAI is following a similar pattern, with new applications now hitting the market on what feels like a daily basis. Some are doubtless destined for the quagmire at the bottom of the trough of disillusionment, but others have emerged as exceptionally strong candidates with genuine business value.

Practical applications for GenAI in eDiscovery

When it comes to practical applications for Gen AI in eDiscovery, there are plenty of areas with great potential. That said, for us, two have already emerged as very clear frontrunners, where GenAI would strongly appear to add value in terms of time- and cost-saving productivity enhancements.

Here’s how.

Early Case Assessment

GenAI’s ability to rapidly and cost-effectively distil a large volume of data down to its essence, is a potential gamechanger for early case assessment.

It enables investigators to perform early analysis on a potentially vast document corpus in order to validate and/or expose key fact patterns that will inform the direction of their case – even whether there is a case to answer in the first place. Through natural language questioning and citing evidential material, it can direct the investigator towards the stones to turn over first in the search for deeper clarity, cutting to the chase faster in the quest to reduce overall eDiscovery costs.

Analysis of Court Transcripts

In a more specific application, GenAI also offers a powerful shortcut for analysing court transcripts in order to build a targeted response or defence. This is traditionally a highly manual and arduous task, often with extremely tight timelines (as in arbitration hearings).

Being able to rapidly  analyse, summarise and distil the salient facts from often lengthy proceedings, can provide a valuable competitive edge, enabling more effective – and cost-effective – litigation and dispute resolution.

GenAI vs manual review

Does GenAI deliver better results than manual review? The answer to that varies by application and perspective.

In our experience, the results are comparable. In many cases GenAI may be more precise and consistent than human reviewers operating under pressure. GenAI is also undeniably quicker and more cost effective than manual review, and therefore more responsive to urgent demands.

It does appear to be the case that the bar for GenAI (and AI in general) has been set very high; perhaps higher than for the human equivalent. Clearly if a key piece of information is missed, it could have a dramatic impact on the outcome of a case. But it makes no difference whether it was missed by the GenAI or the sleep-deprived associate working late into the night…  

So GenAI is not a magic wand or a silver bullet. It is not intended to replace human involvement or complete the whole job of eDiscovery. Like all technology, it is a tool and its efficacy should be measured more by the hand of the person wielding it and how its results are fed into the wider eDiscovery process, checks and balances.  It requires human intervention and, ultimately, human review of the suggested output. But it can vastly augment and improve the investigative, assessment and review processes (amongst other eDiscovery workflows).

Wielding GenAI for maximum impact

The full extent of GenAI’s potential in eDiscovery is yet to be revealed. What we’ve seen so far, however, presents a very compelling picture for considering its adoption.

As mentioned, GenAI is a tool, and its handling plays a critical role in its impact. Learning how to wield its capabilities optimally – and in which situations they hold the greatest benefit – is going to play a critical role in unlocking its full potential moving forward.

In this, a partner like Salient can offer invaluable assistance, leveraging our extensive training and specialist insights to help you wield the right tools, at the right time, in the right place.

Get in touch to find out more about unlocking AI’s potential within your organisation to reduce the time, cost, and workload of eDiscovery.