- MEMBER LOG-IN
Using ChatGPT to speed up contract analysis - with Matt Bishop
Matt Bishop’s latest foray into the technology world aligns with his commitment to adapt and lead in a continually evolving construction sector where he constantly seeks to make construction smarter and more efficient.
Co-founder of PRENGUIN, a software platform revolutionising the way commercial interiors are designed and regulated, and principal at Brevity, a specialist firm in seismic design, compliance and management for interiors, Matt has recently been testing the ability of ChatGPT to add value to the work environment and has come up with the New Zealand Contract Balance Assistant (CBA). This is an artificial intelligence (AI) tool built on the GPT framework, finely tuned to analyse engineering consulting contracts clause by clause.
Why is the New Zealand Contract Balance Assistant a valuable tool?
Contract analysis is time consuming and difficult for those not constantly in the field. It also creates extra work if people need to be trained in how to do it. And, with an increasing number of clients adding new conditions or clauses to contracts, knowing when a contract term is fair is an ongoing battle for engineering consultants trying to navigate these latest contract versions.
Matt created the CBA to provide value for those who experience similar issues in the workplace. He says it is a great first step when analysing a contract because it can alert the user to potential areas of concern, noting the tool is not intended to replace legal advice, which should still be sought as required.
He considers the CBA exemplifies the fast-changing nature of the digital world. Like the acceleration occurring in the ways to use ChatGPT, the CBA accelerates the initial contract analysis process.
How does it work?
Matt says ChatGPT, the famous large language model, is not that powerful when standing alone because it simply predicts words by determining what makes the most sense to say next. However, because language is integral to human existence, it can be powerful when used for a specific purpose. A contract is language-based and, when ChatGPT is provided with the right documents and prompting relevant to a specific contract, it can add value to a company’s operation.
Although ChatGPT appears easy to use, Matt notes much tweaking was required to obtain the desired result. To ensure the CBA works well, a very specific question regarding the balance of each relevant contract clause is required.
Another key ingredient is a foundational contract with which to compare the contract in question. The ACE New Zealand Conditions of Contract for Consulting Services (CCCS) is widely accepted by the engineering and consulting sector as industry standard, and therefore provides a base document to use during analysis.
Matt likens ChatGPT to a conversation. As conversations develop, conversational drift may occur, and the same will happen with the CBA resulting in a departure from the original task if it is not provided with adequate information. Accordingly, rather than simply asking for a comparison of a whole contract to the CCCS, the tool needs constant reminding in the form of questioning at each stage of the clause-by-clause analysis to ensure it remains on track. It also builds a chart of the balance assigned to each clause as it works through the contract. Both the questions and chart serve as constant reminders to stay on task, and demonstrate the subtle prompting required to get good outcomes.
Matt says the tool has a reasonable ability to rate the clause in question according to its own knowledge of what clause balance is like. But, if you ask it to compare the clause to the relevant clause in the CCCS, it will provide feedback by comparing the two clauses and whether the clause in question meets industry standard or not. That often results in a re-distribution of balance given the extra context.
The CBA is a publicly available ChatGPT on the open AI platform and therefore freely available. Despite not currently intending to develop it further, Matt welcomes feedback and says he will continue to work with the CBA if he is made aware of any issues.
Will the CBA be well-received by its target market?
Matt sees the tool being employed by companies seeking an advantage, noting that typical adoption follows the early adopter profile. Early adopters are those willing to take risks because they see the strategic opportunities and possible returns of new technology and will therefore use new products before the majority. These users are also more likely to accept workarounds and higher learning curves as part of the price of being at the leading edge of technology.
Consequently, Matt believes the CBA and tools like these are unlikely to be widely used by most of the relevant market until they are released by an internationally recognised software platform as a business-as-usual product, for example, Microsoft Office or Google Workspace.
Looking to the future – what’s the big picture?
Matt considers the biggest takeaway from this AI experiment is that people need to be working with and aware of these tools as they are released because of their ability to transform the way we work. Otherwise, they risk being left behind.
Discussion is also commencing around digitisation in engineering. To aid understanding of what this may mean, Matt advises replacing the word digitisation with mechanisation and looking back about 125 years. He likens the increasing availability of digital tools to the mechanising of workforces, comparing the digital tools to an engine that is plugged into a workflow to carry out tasks instead of people. This can have wide ranging implications for the consulting industry.
On a positive note, Matt says digitisation enhances the ability to work smarter and more productively if it means removing people from an inefficient process rather than just providing them with smarter tools. And who actually likes reviewing contracts!
Give it a go
You will need a ChatGPTplus account, but you can try out the CBA here.