Middle East | Publication | February 2024
For many reasons, delay and other issues are almost inevitable on a project. There will be claims, and in many projects, these are factually and technically complex, document heavy and involves multiple parties – with different agendas and commercial objectives.
Traditionally, claim consultants are engaged to evaluate, analyse, and manage claims relating to the ultimate trinity of time, cost and quality. Their role is to identify the root cause of disputes, gather evidence, and assess their validity. It seems that there could be significant cost savings if generative AI can be effectively deployed to autonomously generate solutions, designs, or schedules based on data patterns. Despite this, the industry, certainly in the Middle East region seems reluctant to embrace and implement generative AI in the context of claims preparation – are we missing the boat here or is there a valid reason for a cautious approach?
Before we delve deeper, what is generative AI? Generative AI is primarily a data-driven technology and it relies on patterns and quantitative data. It excels at handling large amounts of data especially with process automation, cognitive insight and data analytics. This means it can automate and speed up time-consuming tasks. By analysing resource requirements for a project, generative AI could generate schedules, optimise monthly workflow, and produce monthly reports. Generative AI can also assist with risk mitigation in the pre-dispute stages of a construction project by spotting early warning signs so that the parties could employ mitigation strategies to prevent a potential dispute from arising.
The predictive capabilities of data analytics in generative AI are able to detect variations in a construction project such as changes to scope or specifications. For example, a common dispute trigger between parties stem from revised drawings which may not be entirely complete, clear or accompanied with change orders. By training the generative AI with more data across different data sets, the underlying algorithm is able to record revisions made and avoid the confusion of records between the parties. This can also be used to track consequential increases (or decreases) in costs.
In the e-discovery space, generative AI could further the current tools which are used for generic document review. While current machine learning tools in this space are used to identify and prioritise relevant documents, generative AI could take this a step further by producing summary of keys facts which could be helpful for chronology building at the outset. The real advantage of generative AI in this area is its ability to interpret a query to create meaningful outputs. In essence, it could help to produce responses to prompts based on a specific corpus of documents, which could be very useful in the context of a construction claim.
From our own experience, while the answers produced by the generative AI tools are typically fluent, they are not always factually accurate. Nonetheless, it is a promising technology, and it is expected that future iterations of these tools will be improved to provide citations referencing specific documents to check factual accuracy. It is however important to bear in mind that current forms of generative AI would still be limited in responding to queries with reference to technical and construction related concepts, as its primary set of knowledge is based on the corpus of documents.
Unlike traditional or conversational AI, which is trained from data sets of human input, generative AI is trained on different sets of data to learn patterns to create content with predictive patterns. Essentially it removes the sole reliance on accurate human input to train the AI model, which means that it could possibly be less subjected to human biases. However, there is still a risk that the data sets used to train generative AI which spans across historic cases, data and records may not be kept up-to-date and be ridden with flaws. This would be a problematic consequence as the predictive patterns produced by generative AI could be inaccurate. This is likely to be an issue in this region given that many of the projects on foot or planned are novel, first-of-their-kind projects.
When it comes to the complexities of construction claims, generative AI will struggle with the nuances across both significant particularities which claims consultants are experienced in, namely unstructured data and regulatory frameworks. While there are progressive tools being developed to generate and recognise images, a construction dispute could involve a vast variety of unstructured data such as multimedia recordings and detailed photos of site, it would take an experienced claims consultant to determine the value and relevance of such data in the context of a claim. In relation to regulatory frameworks in the construction sector, there may be specific industry, legal or local specificities which are continually developing. This could have an impact as to how certain issues are being heard from the lens of a judge, mediator or arbitrator, which could also affect the way parties may choose to present their claims. Certainly, in such areas, generative AI is not as developed as human experience in order to consolidate data and to make such accurate judgment calls.
Construction projects are also subject to frequent changes depending on the environment, unforeseen changes in resources, events that occur on-site and emergencies. Here, generative AI is bound to lack real-time knowledge and experience required to tackle this accurately. Generative AI will eventually have to develop the ability to react quickly to these events to remain effective to such issues in the context of rising construction disputes due to environmental and force majeure factors.
Generative AI certainly has a role in the construction sector and the preparations of construction claims – especially given the value and sizes of projects in the region. However, this is not without limitations. The expertise and human element from claim consultants and lawyers in the lifecycle of a construction claim is still essential. However, as technology continues to develop, we should view generative AI and its capabilities as useful tools for us to be more efficient in certain tasks to aid the process of preparing for claims. As a start, parties should invest time and effort at the outset of a project to ensure that they implement the most appropriate technologies to manage project documents and claims – too often archaic technologies are being deployed which ultimately leads to greater costs for all parties in the long run.
Publication
Miranda Cole, Julien Haverals and Emma Clarke of our Brussels/ London offices are the authors of a chapter on procedural issues in merger control that has been published in the third edition of the Global Competition Review’s The Guide to Life Sciences. This covers a number of significant procedural developments that have affected merger review of life sciences transactions.
Subscribe and stay up to date with the latest legal news, information and events . . .
© Norton Rose Fulbright LLP 2023
