This month, we learnt that generative AI, or GenAI, can build an entire website from a hand-drawn diagram, troubleshoot for technicians in the field and win ‘Where’s Wally’ in seconds.
The use cases for GenAI are evolving so rapidly that what is fresh today will be yesterday’s news by the time you read this.
We have entered a new epoch – our own Cambrian explosion. But this time the proliferation is not of new species, but of innovative applications that spark new ideas.
McKinsey, which hailed 2023 as GenAI’s ‘breakout year’, estimates that the technology could add US$4.4 trillion annually to the global economy. IDC forecasts that enterprise spending on GenAI services, software and infrastructure will grow from US$16 billion in 2023 to $143 billion in 2027.
A year on from ChatGPT’s launch, there are now thousands of tools available. Many workers are using GenAI for research, to generate or iterate ideas and write emails, and Deloitte estimates that daily users of GenAI are saving around 5.3 hours each week.
Dozens of ‘business-ready’ Gen AI use cases already apply to construction: site plan optimisation, material design review, virtual field assistance, cost analysis and more. And 55% of businesses, globally, are either piloting or producing GenAI solutions, according to Gartner.
Large Language Models (LLMs) like ChatGPT, Luminous and Llama, put the power of GenAI into everyone’s hands. At RLB, we are experimenting in the AI playground with various LLMs, using our treasure trove of data to run queries and extract fresh insights to age-old problems.
Deep dive into data
RLB’s position of strength is our global reach and our deep reserves of data that goes back decades. We developed our digital system, ROSS 5D, well ahead of the curve and have been innovating with data for many years now.
RLB’s team is focused on GenAI initiatives that support innovation and unearth value that has, until now, been hidden or out of reach. We are investigating data warehouse options to aggregate disparate sources of anonymised data – from cost estimates to building information models – into one accessible and digestible repository. This processed data can be fed into an LLM and our quantity surveyors will be able to ask a chatbot questions. This will catalyse our own Cambrian explosion.
We are also using machine learning (ML) – statistical models that can infer patterns or trends to make predictions – that allow us to ask a multitude of ‘what if?’ questions. After the ML model determines the pattern, an LLM provides a human-centric way to query the results. Essentially, the quantity surveyor chats to the LLM that has ingested the ML’s results. The QS can then better monitor and mitigate potential risks, such as cost or time overruns.
It sounds obvious, but when GenAI can tell us exactly how many hinges are required for each of the doors in a 50-storey tower, our skilled quantity surveyors have extra time to spend value-adding.
Importantly, we will still be required to review and verify data, identify inconsistencies and find efficiencies. But ahead of us lies the promise of greater efficiency and the possibility of more time to spend on high value tasks that only thinking, feeling human beings can execute.
Possibilities and pitfalls
Working with GenAI does come with risks, although eight in 10 business leaders believe that the positives outweigh the pitfalls. Hallucinations – when LLMs generate convincing but inaccurate information – are a new challenge and a type of information error we haven’t encountered before. LLMs are designed to please the person asking the question, even if the answer is wrong, but we are learning that these hallucinations are usually the result of dataset deficiencies.
Another obvious area of concern is data privacy. LLMs are only powerful because they’ve been trained on the creative output of human beings; a recent analysis found 170,000 pirated and copyrighted books were the source of one tech titan’s dataset. Employees are also uploading confidential information into ChatGPT without considering how that data may be used.
RLB is also looking to develop our own proprietary applications using the power of enterprise LLMs so we can protect our data and that of our clients. We have also established a responsible data processing policy that defines principles of fairness, transparency and accountability.
No job role will be untouched by GenAI, and Goldman Sachs analysts have predicted that more than 300 million of today’s jobs will be automated. But technology has traditionally created more jobs than it has displaced. According to the World Economic Forum’s Future of Jobs Report 2023 the impact over the next five years will be net positive.
Rethinking the way we think
GenAI will also require all of us to rethink the way we think. One illuminating research project undertaken by Boston Consulting Group found a 40% improvement in performance when skilled professionals used GenAI for creative ideation. But when the same skilled professionals were asked to undertake business problem solving – a capability outside GenAI’s current frontier but well within their own remit – it led to a 23% decrease in performance.
What this tells us is that GenAI can identify unexpected, and sometimes even counterintuitive, patterns and correlations that can spark new ideas. But human intelligence determines how we use these ideas.
Today’s quantity surveyor requires a different set of skills to those we have relied on in the past. Our role is no longer limited to preparing cost estimates, but to harness data and unlock insights that create new value for clients. As we do this, quantity surveyors will move up the value curve.
When ChatGPT launched to the public in November 2022 it was like watching Roger Bannister break the four-minute mile barrier. Now we know it can be done. We understand that GenAI will change everything – we just don’t know how yet. But just as the Cambrian Explosion transformed the Earth’s biodiversity, GenAI is reshaping the digital ecosystem and ushering in a new era of boundless possibilities.
John Corry
