Do AI Assisted Litigation in 2024

January 2024

The generative artificial intelligence (AI) advancements of 2023 are paving the way for an array of new software features in 2024. This article delves into some of the eDiscovery AI services and solutions that are available to lawyers now and outlines strategies for effectively adopting this burgeoning technology.

“Not another generative AI hype piece,” you may be thinking. I empathise with the criticism. Rest assured, this article won't stray into typical frustrations when discussing AI. It will not muddle generative AI with commonplace machine learning tasks, nor will it sensationalise the implications of advanced AI for clicks. This post does not attempt to predict the impending end of the legal profession. Nor does it attempt to cover all scenarios where generative AI might be useful. Instead, the post aims to spotlight some of the practical generative AI services available now and how they can be adopted in litigation.  

In the past year, AI has been a hot topic around the firm water cooler. Discussions range from "Did you hear about the lawyer who cited non-existent cases using ChatGPT?" and “What about the copyright infringement allegations against OpenAI?” As AI continues to make headlines, both positive and negative, being an informed adopter is crucial to avoid pitfalls.

Despite public missteps, 2024 will see continued integration and adoption of generative AI in litigation. Generative AI has evolved beyond a “tech demonstration” to offer tangible value to lawyers (among others). In less than a year, impactful generative AI services have become programmatically available, and we are witnessing the integration of AI with our everyday tools and services.

Reflecting on the adoption process for automation technologies in other sectors can be instructive. For example, the Society of Automotive Engineers’ six-stage process of evolving the driving experience from manual to fully automated illustrates a gradual shift in responsibility from human to AI.¹ This model – keeping humans "in the loop" until AI services mature – seems a sensible approach for litigators as well.

We are at a pivotal moment with generative AI. OpenAI's GPT-4, developed in 2022 with over a trillion inputs (estimated) and ten times more than its predecessor, is a testament to this.² The competitive field is heating up also, with major players like Google entering the fray and Amazon investing in OpenAI alternative, Anthropic.³ This competition will undoubtedly fuel further innovation.

Interestingly, jurisdictional competition is also emerging. Some countries are attempting to legislatively deal with copyright concerns stemming from training data, while others advocate for “comprehensive and distributed AI regulation.”⁴ Novel concepts like floating self-sovereign data centres, immune to EU or US regulation, are being canvased.⁵ The outcomes of these varied regulatory efforts to control AI innovation will be intriguing to observe.

However, do not be deterred by the novelty of the technology and its rapid pace of change. Generative AI offers innovative, productivity-enhancing solutions for litigation today. Its benefits, costs, and imperfections are all factors to weigh for its use on specific projects and tasks.

Generative AI offers innovative, productivity-enhancing solutions for litigation today. Its benefits, costs, and imperfections are all factors to weigh for its use on specific projects and tasks.

To mitigate the impact of imperfections, focus on tasks that are inefficient to manually undertake, deal with discrete data volumes, and where the anticipated result has a clear benefit. Seek out simpler tasks where AI results do not need to be perfect, but directionally correct results can be more beneficial than existing alternatives.

In the last three months, we (Adio) have identified several tasks with our clients that can benefit from generative AI, including:

• Summarising and identifying key topics in lengthy documents for initial issue-related review.
• Providing specific answers about a document's relevant content and its location.
• Highlighting dates and figures, variably formatted and challenging to locate via keyword searches.
• Automating the creation of document descriptions for chronologies.
• Analysing and synthesising narratives across sets of relevant documents.
• Extracting objective information from the face of documents to prepare lists for courts.

Using AI is not without risks. Data sovereignty, output accuracy, inherent training biases, and potential hallucinations are all considerations. Service providers and software vendors will need to be transparent about the use cases, deployment methods, and imperfections helps facilitate mature discussions about risk management.

There are many operational considerations, including task selection for AI integration, crafting effective prompts, algorithm choice (not all AI is equal, far from it), managing context window limitations, and scaling solutions to avoid cost overruns and unsuccessful outcomes. Getting familiar with how the technology operates and how you can programmatically interact with it will further your understanding of the benefits and limitations of AI.

Despite these challenges, the potential for generative AI in litigation is immense. Service providers can lead the way in exploring innovative, time-saving AI solutions. Our company, Adio, is currently exploring use cases involving one-time document enrichment and limited/staged document processing with AI. The results of using AI to enrich documents have been impressive and useful.

The results of using AI to enrich documents have been impressive and useful.

The past year has seen significant advancements in generative AI, and its integration into eDiscovery is just beginning. Efforts to understand how this technology can support lawyers today are far from futile. The quality and value of AI-enhanced services will only improve with time and feedback from legal teams. Accordingly, we are excited to collaborate with litigation teams this year in safely harnessing this innovative technology.

Ben Kennedy
Managing Director, Adio

¹ https://www.synopsys.com/automotive/autonomous-driving-levels.html.

² https://en.wikipedia.org/wiki/GPT-4.

³ Google are releasing their generative AI solutions in 2024 https://deepmind.google/technologies/gemini/#capabilities and Amazon invest $4 billion into AI company, Anthropic https://www.forbes.com/sites/stevendickens/2023/09/26/amazon-invests-4-billion-in-anthropic-a-paradigm-shift-in-ai.

⁴ Article 30-4, Copyright Act of Japan (Act No. 48 of 1970). The unofficial version is available at https://www.japaneselawtranslation.go.jp/ja/laws/view/4207. Brookings’s proposal for regulation is available at https://www.brookings.edu/articles/a-comprehensive-and-distributed-approach-to-ai-regulation.

⁵ Floating datacentre with 10,000 graphical processing units specifically for AI use https://www.delcomplex.com/blue-sea-frontier. The European Union is drafting an AI Act in an attempt to regulate the development and use of the technology https://www.cliffordchance.com/insights/resources/blogs/talking-tech/en/articles/2023/04/the-eu-ai-act--concerns-and-criticism.html.