Skip to main content
Bulletin

Does Not Compute: ChatGPT and the Evolution of Legal Practice

Fasken
Reading Time 5 minute read
Subscribe
Share
  • LinkedIn

Overview

Technology, Media & Telecommunications Bulletin

November 2022 saw the public launch of artificial intelligence (“AI”) tool ChatGPT, by AI research and development company OpenAI. ChatGPT’s rise to prominence has been swift, reaching the million user mark in just five days. ChatGPT is an AI chatbot that can generate coherent, novel responses, drawing on a vast dataset of written works from the internet. ChatGPT utilises a natural language generation model, meaning that the AI underpinning the chatbot will comb through huge amounts of data, not only processing the data but also assessing the relationships between words and how words come in sequences. Based on its assessment, the AI will then generate responses to prompts and questions. This enables ChatGPT to formulate coherent and at times detailed responses, learning from the human feedback it receives. While the technology is still in its relative infancy the potential applications are huge and the legal sector has been taking note. 

Looking at the use case for ChatGPT in the legal sector, lawyers are under a constant pressure to provide expeditious, cost-efficient and commercially minded services. In the corporate world, over the lifetime of a transaction huge proportions of time are devoted to document analysis, research, due diligence and correspondence. Subject to ChatGPT being granted access to a broader data set, ChatGPT will be able to automate such tasks. The AI tool provides users the ability to review documents, analyse data / document patterns, highlight key pieces of information and draft email summaries to send onto clients in moments. While this will certainly need to be checked over and there will inevitably be mistakes and key issues missed, the potential to streamline workstreams and decrease the time and money spent on document review, research and correspondence is clear. To this end we have already seen law firms roll out equivalent tools, such as ‘Harvey’ (built using OpenAI technologies), a chatbot intended to help lawyers with a variety of due diligence and document review related tasks.

In addition to this, ChatGPT is also able to produce first drafts of legal documentation and contracts in seconds. This could be incredibly useful for typically shorter documents such as non disclosure agreements and wills, as opposed to longer form documents such as share purchase agreements. Further to this, as ChatGPT is provided with more information the documents it produces will start to become more legally compliant, meaning over time the errors and issues with legal documentation should start to diminish. ChatGPT could also be a useful tool when it comes to drafting clauses in isolation, in accordance with specific instructions. When given the right instructions and prompts, ChatGPT could generate a clause in moments which can be edited and further tailored to any contract or agreement. This could again save lawyers huge amounts of drafting time, with the technology continuously improving and learning through each input.

However, there are issues to be mindful of when looking to utilise the technology. One of the key concerns being AI tools are only as good as the data they utilise, with faulty data resulting in incorrect or inconsistent responses. Further to this, ChatGPT only has access to data before 2021. Consequently, the AI may be prone to bias or inconsistencies depending on the data it is provided with, meaning that all responses will be in some ways altered by the bias and potentially incorrect. Following on from this, intellectual property (“IP”) infringement may also be a big issue. If data, specifically textual data, is fed into the AI and that data attracts intellectual property rights, it is a foreseeable possibility that the AI may use the protected data as a basis for their responses, passing it off as novel or original, thus inadvertently breaching IP laws, potentially leading to legal action if discovered. While arguments can be made around fair use, it is still a serious concern.

A further key issue to consider concerning ChatGPT is that of data protection. Currently, OpenAI does not offer any means for individuals to delete their personal information or check if and how OpenAI stores their personal information. This is in breach of the General Data Protection Regulation (“GDPR”) and UK GDPR and could cause issues if UK or European businesses wanted to utilise the tool in their everyday practices. Additionally, OpenAI is yet to disclose its data collecting method, which again causes UK GDPR concerns. Thus, when using ChatGPT, users need to be mindful that there is a risk of their personal information being incorrectly handled, with users having no access to, or control over, their data, contrary to data protection principles. The current data protection issues make the implementation of ChatGPT in UK / European businesses unlikely as things currently stand.

AI in the legal sector is not a new concept and has been around for years. Law firms have often sought to utilise AI in order to evaluate the probability and profitability of civil matters, assist with legal research and undertake high level document reviews. ChatGPT while the latest iteration of AI, proposes a real value prospect for the legal industry. ChatGPT offers the ability for lawyers to streamline their days focussing only on the most important and complex matters. However, there are a number of serious issues and barriers to implementation that need to be overcome. Until OpenAI is able to solve their data protection and IP related issues there is no way law firms can use the technology in its current iteration. Despite this, the technology is exciting and the legal world is seriously looking at methods of implementation, even going so far as developing their own compliant AI systems.

    Subscribe

    Receive email updates from our team

    Subscribe