AI (Artificial Intelligence) and more generally data science (which combines AI, machine learning, advanced analytics, maths, and statistics) are rapidly developing technologies that are becoming increasingly accessible. Most of us carry a Siri or Alexa around with us on our devices, and in late November 2022 OpenAI made its ‘AI language tool’ ChatGPT freely available to the mainstream.
Author: Ibrahim Elsisi, Associate Director, Brisbane, Australia
A study for the Swiss Bank UBS reported ChatGPT to be the fastest growing app in history [1], reaching 100 million monthly active users just two months after its launch. For comparison TikTok took nine months to hit that uptake; Instagram, two and a half years.
In response to the prompt ‘what are the three biggest strengths of AI?’ ChatGPT tells us:
“1. AI excels at processing and analysing vast amounts of data quickly and efficiently, enabling it to identify patterns, make predictions, and uncover insights that humans may not easily detect.
2. AI has the ability to automate repetitive and mundane tasks, freeing up human resources and allowing them to focus on more complex and creative endeavours.
3. AI can assist in solving complex problems by considering numerous variables simultaneously, providing optimized solutions and recommendations that can enhance decision-making processes. (sic)”
These strengths mean that AI and data science technologies offer immense potential to construction professionals. The construction industry is renowned for its complexity and scale, as are the disputes.
Construction professionals face a dual challenge; the need for complex analysis on extensive documents and technical data, balanced with the requirement to present the results in a way that can be understood and practically applied. That is why construction professionals are turning to data science and, in more recent times, AI. Its strengths offer the potential to enhance efficiency and accuracy.
Automated document analysis: streamlining information extraction
Construction projects generate extensive quantities of data and documents. This includes contracts, correspondence, technical details and performance records. Traditional manual analysis of this data can be a laborious and error-prone task which is limited by cost driven predefined criteria.
AI-powered document analysis, Natural Language Processing (NLP) techniques and training AI models on large datasets of construction-related documents can automate this process, based on much larger predefined criteria or keywords given the overall lower costs when compared to manual analysis. For example, AI can be trained to automatically extract information related to specific project delays or variations. This automation saves time and equivalent costs, while freeing professionals to focus on thought-orientated tasks such as analysing and explaining the extracted information. Similarly, AI techniques can be applied to quickly identify discrepancies or inconsistencies between different documents (such as numerous revisions of a particular general arrangement drawing), identifying areas for deeper investigation.
Productivity trend identification: analysing past performance and forecasting future performance
Data science techniques provide construction professionals with insights into performance on a construction project and productivity trends. Regression analysis and machine learning algorithms can be applied to analyse the relationship between known project variables such as duration, resource allocation, cost, and productivity. Machine learning models trained on historical data can also provide predictions and forecasts for future performance, allowing professionals to anticipate risks and opportunities.
Techniques such as decision trees, random forests, and neural networks can quickly analyse various project factors and their impact on performance. These techniques help identify both short and longer term patterns and correlations, enabling professionals to make informed decisions and provide data-supported conclusions.
Scenario-based delay analysis: objective evaluation in complex disputes
Experience dictates that disputes arise from differences of opinion which are not necessarily based on fact. Data science and AI can play a crucial role in providing objective evaluations. For instance, by training AI models on historical project data and applying machine learning algorithms, professionals can simulate and evaluate different analysis methods based on the available data. A common application is the analysis of delays. AI models can analyze project schedules, performance data and other available details to assess the effect of disruptive events. By automating these analyses, professionals can reduce manual effort whilst providing evidence-based insights. The ability to run multiple scenarios offers a more objective analysis by reducing human bias and subjective preferences for certain approaches.
Simplifying technical findings: effective communication for non- technical parties
AI tools, when used appropriately, simplify complex technical information, aiding effective understanding and communication between stakeholders with varying technical skills. Decision tree algorithms provide step-by-step guidance through complex decision- making processes. Additionally, AI language tools such as ChatGPT can be prompted to generate plain language explanations and summaries. For example,
ChatGPT’s one-sentence summary of this paragraph is: “AI tools simplify complex technical information, guide decision-making, and generate clear explanations for effective communication." By simplifying complex technical language, AI tools enhance understanding, transparency, and inclusivity for all parties involved.
Cautionary considerations: limitations and human expertise
While data science and AI offer significant advantages in construction data analysis, it is important to be aware of their limitations and exercise caution in their use.
AI’s creative language limitations have been famously highlighted by Nick Cave [2] among others, and reliance on ChatGPT for legal research has recently landed a US lawyer in difficulty [3]. AI models can be influenced by biases in the training data, potentially leading to inaccuracies. Complex or ambiguous language may challenge AI's document analysis capabilities, risking misapplication of the nuances and context of construction-related documents. A leading criticism of nascent AI is that if it makes an error it often remains committed to that error, continuing to believe that it is right. Human professional expertise and judgment remain essential in verifying results and ensuring appropriate application.
The key to the successful use of data science and AI is the availability of relevant, reliable and consistent data. There is an old adage of “garbage in – garbage out” and the professionals need to be clear on the data being used to support the ensuing results.
Conclusion
Data science and AI integration have the potential to revolutionize the analysis of construction project data, particularly for dispute resolution. Automated document analysis streamlines information extraction, while productivity trend identification enables data- supported decision-making. Multiple scenario testing, objective evaluations, and simplification of technical findings promotes effective communication and informed, inclusive decision-making. Balancing AI’s evolving capabilities with human professional expertise ensures effective data-supported outcomes, ultimately benefiting collaboration, reducing costs, and improving the construction industry.
Finally, the reliance on data science and AI has similarities to the reliance on a complex time schedule. If the output is 99% correct, the 1% that is incorrect can undermine confidence in the entire results.
Going back to the difficulties of the US lawyer, the BBC reported that the ChatGPT prepared filing referred to an example case that did not exist. It is likely that this was a very difficult position from which to recover in the eyes of the judge.
This article was written for issue 27 of the Diales Digest. To view the publication, please visit: www.diales.com/diales-digest-issue-27
1. ‘ChatGPT sets record for fastest-growing user base’; February 2, 2023 - https://www.reuters.com/technology/chatgpt-sets-record-fastest-growing-user-base-analyst-note-2023-02-01/
2. ‘Nick Cave Slams AI Attempts at Nick Cave Songs’; January 16, 2023 - https://www.rollingstone.com/music/music-news/ai-chatbot-chatgpt-writes-song-nick-cave-style-1234661842/
3. ‘ChatGPT: US lawyer admits using AI for case research’; 27 May 2023; A New York lawyer is facing a court hearing of his own after his firm used AI tool ChatGPT for legal research - https://www.bbc.com/news/world-us-canada-65735769