News

Industries

Companies

Jobs

Events

People

Video

Audio

Galleries

My Biz

Submit content

My Account

Advertise with us

Law Practice News South Africa

Pros and cons of gen AI in the legal sector

For the past few years, generative artificial intelligence (gen AI) has become the buzzword across most industries, including health care, financial services, and recently legal. Gen AI is a subcategory of artificial intelligence that is capable of producing text, images, video and audio. An example of gen AI is ChatGPT – a gen AI chatbot where users can ask it questions and it will generate human-like responses.
Image source: bpawesome –
Image source: bpawesome – 123RF.com

If you are a millennial or Gen Z, you are likely to have embraced gen AI and view it as an essential tool for improving your efficiency and productivity. Other generations remain sceptical of using gen AI tools due to their propensity to produce inaccurate results and potentially misuse personal data.

While it may be difficult to persuade certain people to embrace gen AI, potentially the ultimate challenge is convincing attorneys to do so. Firstly, the legal profession is generally slow to adopt new technologies; and secondly, the legal profession is the antithesis of gen AI tools in that it is characterised by trust and certainty, driven by precedent, and emphatic about maintaining confidentiality and legal privilege.

I understand why attorneys are apprehensive of using gen AI tools. Although these tools might reduce the amount of time spent on routine tasks, such as legal research, the risk of client’s confidential or legally privileged information being disclosed by using a gen AI tool outweighs any time-saving benefit. Similarly, it seems non-sensical to use a tool for legal work that has issues with accuracy and bias, as well as being prone to “hallucinations” (a response generated by AI that contains false or misleading information, which is presented as fact).

If, however, these issues are resolved, which “AI legal assistants” such as CoCounsel and Harvey AI claim to have done, it is possible that the notion of an AI legal assistant will become more palatable to attorneys.

Examples of existing gen AI tools for the legal profession

One gen AI legal assistant that has gained traction and already been implemented at several global law firms, such DLA Piper, is Thomson Reuters’ aptly named gen AI assistant, “CoCounsel”. Thomson Reuters refers to CoCounsel as “the first and only professional-grade GenAI assistant” and “the only partner you can rely on, wherever you’re working”.

Some of CoCounsel’s capabilities include reviewing up to 1000 documents in minutes, interpreting and condensing critical information in any type of document, and my favourite – creating event timelines.

In addition, CoCounsel claims to be able to read, comprehend and write at a postgraduate level and complete the work of a junior associate in minutes. With these capabilities it is not surprising that CoCounsel is being used by attorneys to improve the services they offer to clients.

If CoCounsel is as good as it claims to be, it seems to create an opportunity for law firms (who can afford to purchase such gen AI tools) to reduce the time and costs associated with tasks such as research, summaries, and document reviews, and offer legal services to clients in a cost-effective manner. It also potentially levels the playing field between small and large firms by allowing small firms access to time and cost-saving resources.

Regardless of how effective an AI legal assistant might be, its appeal and longevity will be dependent on its accuracy and ability to keep client’s confidential and legally privileged information secure at all times.

Melissa Steele is a Senior Associate at boutique law firm, Nortons Inc.
Melissa Steele is a Senior Associate at boutique law firm, Nortons Inc.

The accuracy of Gen AI tools

Gen AI tools are reliant on the quality of their training data and any inaccuracies in data can severely impact their performance. In other words, if a gen AI model is trained on biased or incorrect data, it will produce biased and inaccurate results. This is why it is critical for gen AI models to have access to quality data.

CoCounsel claims that attorneys can rely on it because it uses the right information, such as GPT-4 and its proprietary search technology, Parallel Search. CoCounsel claims that it does not “hallucinate” due to it “implementing controls to limit CoCounsel to answering from known, reliable data sources such as our comprehensive, up-to-date database of case law, statutes, regulations, and codes.”

However, CoCounsel caveats its claims of accuracy by stating that it was not designed to replace attorneys, and it remains necessary for attorneys to verify the program’s output.

The importance of having access to quality legal data to improve the performance of gen AI models is shown in Harvey AI’s recent interest in the legal research company, vLex. Harvey is a legal AI startup offering gen AI products to law firms, which can analyse and interpret documents; summarise and draft legal documents, such as contracts; record billable hours; and answer research questions.

It is understood that Harvey sought to acquire vLex to use its database of regulatory and legal documents, which would allow it to compete with one of its main competitors, CoCounsel.

Privacy and security risks

Potentially the greatest risk with using gen AI tools is breaching client confidentiality or waiving legal privilege. If an attorney uses a gen AI tool in their practice it will need to ensure that the tool does not retain, store or collect client’s confidential information or share this information with unauthorised third parties.

As I mentioned above, gen AI is trained on the information that is provided to it, which means that there may be a risk of client’s data being used to train the gen AI tool.

Certain gen AI tools provide assurances that client’s data will remain secure and not be used for training purposes. For example, CoCounsel states that it “uses dedicated servers to access the underlying LLMs, meaning your data isn’t sent to ‘train’ the model as part of publicly accessible knowledge.”

Notwithstanding these assurances, attorneys should be mindful of the risks associated with using gen AI tools, including the potential for client’s confidential information being disclosed or legal privilege being waived.

If you or your law firm is considering using gen AI tools, I suggest the following:

  • Obtain your client’s consent prior to using gen AI tools or so-called “AI legal assistants”;
  • Consider incorporating provisions in confidentiality agreements that contemplate the use of gen-AI tools, particularly in relation to discovery proceedings where confidential documents are shared with opposing attorneys who may upload your client’s confidential documents to gen-AI platforms;
  • Read the gen-AI tool’s policy on privacy and data security before using the tool;
  • Opt-out of gen AI tools retaining, storing or sharing your and your client’s data; and
  • Always verify that the information that the gen AI tool has provided is accurate.

About Melissa Steele

Melissa Steele is a Senior Associate at boutique law firm, Nortons Inc.
Let's do Biz