Generative AI's future in law hinges on trustworthy knowledge repositories
Generative artificial intelligence (AI) will shape the future of the legal sector, but its application will be dependent on a trustworthy knowledge repository, says Carlos Garca-Egocheaga, Chief Executive Officer at Lexsoft Systems. The functionality of generative AI will be underpinned by its ability to process natural language on provided data to deliver precise results and outcomes, however, the AI application does not comprehend legal concepts in the questions it answers nor does it understand the words or output it generates. Consequently, the CEO emphasised that the use of this technology requires expertise within the legal sector to validate the outputs the generative AI tool produces.
The generative AI tool has its intrinsic value and will be beneficial to those who understand its underlying mechanics. It can aid in refining thought processes to reach the desired goal. The same query can be issued in various ways, with the AI tool offering different responses. Lawyers can receive varying outputs to identical questions, depending on the documents they have authorisation to access.
Microsoft's role in facilitating the adoption of generative AI, including compliance and confidentiality, will be pivotal. Legal professionals utilising generative AI to prepare reports for clients or third parties will need to ensure compliance with data protection and data privacy regulations. Microsoft's Azure security policy guarantees that data residing in individual organisations' cloud tenants are not externally shared. Therefore, law firms will only be permitted to use data within their own cloud tenants, as the case was when data resided within companies' networks.
Data confidentiality will also pose an obstacle. Individuals will only have access to documents they are authorised to view according to the principles of need-to-know security. The generative AI tool will only generate results based on data the user is permitted to access. The output may vary from individual to individual, potentially impacting the value of generative AI in terms of contextual accuracy, currency, and suitability.
Document Management Systems are the logical place for law firms to embark on their AI journey, and to ensure the AI tool is trained on accurate, authorised, and up-to-date data, firms need to create a central, trustworthy repository of data through the knowledge management system. A lawyer could instruct the generative AI tool to generate an 800 words abstract based on a 100-page merger and acquisition contract involving participants from the US and Germany, and relating to the jurisdiction of New York - and the lawyer can be assured that the correct data was used to create the output.
Nevertheless, even with the potential of generative AI, the legal sector is far from experiencing a 'rise of the machines,' alluded Mr Garca-Egocheaga. Among other things, there are several hiccups to iron out - including AI hallucinations, security, ethics, data management, and regulation - implying that human intellect remains crucial. Generative AI engines process language sequentially but lack understanding of the words or concepts it generates and requires detailed instruction and high-quality data to deliver value.
