By Josh Levy
It is well documented that the construction Industry is way behind the curve when it comes to adopting and embracing technological advancement. Simply put, other major industries are earlier adopters of tech trends.
The construction industry’s cultural nuances and other underpinnings which explain the “why” behind this phenomenon would be a fascinating topic to explore; and one that we would endeavour to address if back in graduate school or pursuing a PHD in psychology. However, and while still glacial compared to the overall constellation of industries, the pace of adoption and acceptance is beginning to accelerate as the construction industry becomes better educated, millennials (and even more tech-savvy generations) are entering management/leadership functions, and more practical applications of tech are becoming obvious and apparent.
To this end there is much being discussed about how AI (Artificial Intelligence) can help the industry. While AI – which is a broad term — has numerous functions in construction, the focus of this article is on utilizing AI to extract data and key risk factors which exist in the voluminous documents which govern construction projects, companies, underwriting decisions, and, ultimately, balance sheets of all industry stakeholders.
The basis for why this use case for AI has become topical in construction is directly correlated to the nature of our industry: big buildings are built upon MOUNTAINS of paper. Yet the people managing the day to day want to build buildings, and have little to no appetite for boilerplates and legalese.
Think about the administrative burden on mid to small sized construction companies, or the 20-something project manager for a large company: to manage a typical construction operation means that one must administrate through large, intricate contracts, subcontracts, scopes, and specifications. On top of that, these documents are almost always intertwined with accompanying insurance policies. That is a lot of fine print! And those fortunate soles who can make sense of this web of risk are usually only comprised of expensive lawyers and/or sophisticated insurance brokers/risk managers.
Which is why many blue-chip construction companies have invested substantial resources into building out in house legal departments, insurance compliance functions, and other related bodies of oversight and governance. While that is certainly a robust approach, most of our industry (including subcontractors) simply do not have the volume or profitability to justify making such a large overhead investment.
Therefore in the vast majority of our huge industry, core operators are having to seek out others’ help to manage terms which govern the projects that must be built. As one can imagine, managing all of this paperwork in real-time is no easy feat and having technology to assist in that is a big help.
And that is why savvy companies, both big and small, are starting to use “Artificial Intelligence” and “Machine Learning” to help de-risk their operations: companies and people are able to pull the AI lever (which is much more cost effective than relying upon humans) to gain critical insights into their construction documents which have previously gone either unappreciated (until something bad happened), or only realized upon engagement of professionals (e.g., does my contract have a waiver of subrogation, how are notices perfected, what are considered delay events, etc.). And blue-chip construction companies are now able to use AI to extract historical data and information which allows data driven decisions to be made.
Well, what is Artificial Intelligence? What is Machine Learning? AI is a broad concept relating to computers or machines performing tasks that simulate human behaviour such as reviewing documents. Machine learning is a subset of that, allowing that same computer or machine to automatically learn from its prior review of data, without having to be expressly updated or programmed for each case. This was possible because AI systems are able to “learn” how to automatically identify, tag and extract relevant items the way a lawyer, risk manager, or savvy executive would. A human will still need to review the product of the AI, but the time saved in avoiding to have to painstakingly find needles in haystacks proves its worth. The same can be done with contract and policy reviews.
AI technology using Natural Language Processing can be trained to identify and tag phrases in contracts and insurance policies (among other document types), directing the reviewer to the relevant provisions and language. These systems are a great way to conduct a “first pass” through a document for speed, as well as a “final pass” for accuracy, to confirm that nothing important was missed or changed before being executed. These systems can also be used to identify historically – hello underwriters!– what types of risks a company has taken on, how those risks correlate to the company’s results, etc. In other words, AI is an extremely powerful tool to extract big data to help inform future decision making.
The point for those reading this is that our industry must continue to evolve as new cost-effective risk management tools come online. It doesn’t make sense (and in fact could be extremely detrimental) to continue operating same as normal, when the barrier to becoming better and more risk-proactive is lower than it has ever been… because of technology. The above described use-case for AI is the most recent example.
Josh Levy, based in Florida, is co-founder and strategic advisor, Document Crunch and director and contracts lead, process and energy at Wood. Oil and Gas.