How to Harness the Untapped Potential of AI in International Criminal Law


TL;DR Breakdown

  • Artificial Intelligence (AI) can revolutionize international criminal law (ICL) if its limitations are understood and harnessed effectively.
  • Complex documents, diverse jurisprudence, and data security risks pose hurdles to AI adoption in ICL.
  • Summarizing jurisprudence, legal research, drafting, evidentiary research, and legal analytics offer promising avenues for AI integration.

Artificial Intelligence (AI), especially large language models (LLMs) like OpenAI’s ChatGPT, Bing Chat, and Google Bard, has raised concerns when used by lawyers without understanding their limitations. Recent cases have highlighted the risks of relying on AI-generated content without proper verification. However, the transformative potential of these tools in the legal profession should not be overlooked. Numerous legal technology start-ups and scholarly contributions have emerged, showcasing AI’s possibilities. In international criminal law (ICL), using LLMs remains largely untapped. The extent to which these tools reshape ICL will depend on their ability to navigate the unique features of this field.

Particular features of international criminal law

ICL litigation heavily relies on complex documents. Investigating international crimes often leads to a vast amount of documentary evidence. While LLMs are designed to comprehend large volumes of data, challenges arise when dealing with poor-quality scans and documents in less commonly used languages not extensively covered in their training.

Furthermore, ICL jurisprudence is diverse, with each institution having unique repositories for storing legal precedents. There are no publicly available AI tools specifically trained on these compilations, unlike domestic legal practice, where American lawyers can easily integrate AI through platforms like Westlaw Edge, Lexis+, and Casetext.

Data security risks pose another challenge in ICL. The security implications of LLMs are still not fully defined, creating potential threats. Data breaches in the prosecution of war crimes or crimes against humanity could have severe consequences, including the identification and targeting of victims, witnesses, and others at risk.

The present and future of LLMs in ICL

Despite the challenges, LLMs already offer significant benefits to ICL lawyers, and their potential extends beyond the distant future. However, proper understanding and application of these models are crucial (refer to the book-length survey here for a comprehensive overview).

Using LLMs requires a paradigm shift from traditional search engine queries and Boolean search terms. Lawyers need to engage in dialogue with these models and craft prompts adeptly to enhance the quality of responses. An upcoming article by Daniel Schwartz and Jonathan H. Choi provides valuable insights into how to prompt LLMs in legal contexts.

It is important to note that while I am an ICL-trained lawyer, I am more of an enthusiast than a skilled user of these tools. More tech-savvy individuals can provide further refinement of my description of LLMs and their potential. For a deeper understanding of LLM development, consult Andrej Karpathy’s video presentation on OpenAI from May 2023.

Applications of LLMs in ICL

Jurisprudence/Filing summaries

LLMs can provide immediate assistance in summarizing jurisprudence or specific parts thereof. The quality of the summary depends on the formulation of the prompting questions. Lawyers can obtain higher-quality summaries by providing relevant passages from judgments and asking LLMs to summarize them. Quoting the text allows for meaningful follow-ups with the LLM, seeking answers with direct quotes from the provided passage.

However, character limits pose an obstacle to further growth. Current subscription-based LLMs have limitations, such as ChatGPT4’s 4000-character limit and Bing Chat’s 2000-character limit. ICL judgments often exceed these limits. GPT add-ons and tools like ChatPdf are being developed to circumvent these constraints.

Legal research

LLMs excel in answering legal research questions that go beyond summarizing known filings. While simpler and widely discussed topics yield more accurate information (e.g., elements of the joint criminal enterprise), more obscure questions may require specific dialogue to obtain meaningful answers. Asking LLMs to develop specific parts of their answers helps uncover the desired information.

Hallucinations are a current obstacle to further growth in this area. LLMs occasionally provide confident but inaccurate citations and even create fake cases. It is crucial to consider LLMs as complements to diligent legal practice rather than substitutes. Additionally, ChatGPT’s training data only extends to the end of 2021, limiting access to the most recent case law (Bing Chat does not face the same issue).

Drafting and editing

Properly prompted with relevant facts and applicable law, LLMs can generate draft paragraphs that are starting points for ICL drafting. They can also enhance already drafted text by improving clarity and readability. Requesting specific writing styles, such as that of “William Schabas” or “an ICC judge,” may yield better results.

One obstacle to further growth in this area is the extent of revisions. Current LLMs cannot generate passages with references suitable for direct use in a draft without further alteration. Proposed changes in word choice can lead to unintended errors when precision is crucial, and LLMs may not always provide clear quotations to explain their generated sentences. Therefore, passages generated by LLMs should be checked and verified before being incorporated into broader work.

Disclosure and evidentiary collection research

LLMs trained on large evidentiary collections can provide valuable insights into the collection. Lawyers can prompt LLMs with information from pending disclosure requests or specific facts relevant to ICL trials. LLMs can assist in summarizing witness testimony, isolating key details, and aiding in evidence capture during trials.

Confidentiality remains an obstacle to further growth in this area. Current technology’s error risks make using sensitive information in publicly available LLMs impractical. Solutions tailored to ICL institutions must inspire sufficient confidence in utilizing confidential data collections. Several tech startups, such as Casetext’s Co-Counsel tool, develop such solutions. The ICC OTP’s evidence submission initiative, incorporating AI and machine learning, represents a welcome development.

 Legal analytics

LLMs’ ability to provide highly specific answers in the face of vast data could revolutionize legal analytics in ICL. By leveraging the right datasets and prompts, LLMs could offer valuable insights into how ICL litigators could respond to certain litigation strategies. Data projections on a judge’s rulings, defense attorney tactics, or sentencing ranges may become possible.

However, the limited number of ICL trials poses an obstacle to meaningful predictions through legal analytics. With frequent turnover among judges, particularly at the ICC, the small dataset size for specific events hinders the development of accurate analytics. Legal analytics may prove more useful for repetitive elements of trials that accumulate larger datasets, such as a judge’s record on questioning objections.

LLMs can upgrade international criminal law practice

The potential of LLMs in ICL is undeniable, even though their current utilization faces challenges. Their rapid evolution makes predicting widespread adoption within the field difficult. ICL can benefit from exploring how to maximize the potential of LLMs while ensuring responsible use. By navigating the unique features of ICL and addressing existing obstacles, LLMs can revolutionize the practice of international criminal law.

Share link:

John Palmer

John Palmer is an enthusiastic crypto writer with an interest in Bitcoin, Blockchain, and technical analysis. With a focus on daily market analysis, his research helps traders and investors alike. His particular interest in digital wallets and blockchain aids his audience.

Most read

Loading Most Read articles...

Stay on top of crypto news, get daily updates in your inbox

Related News

Electric Vehicle
Subscribe to CryptoPolitan