QS v GPT: a fight to the death or a match made in heaven?

There is a widely held assumption that there will always be a requirement for Quantity Surveyors, and most who provide professional services might struggle to see a future where their profession does not feature.

Whilst this might be strong reasons to support that argument, it has been or will be tested by the rapid development and deployment of generative AI, particularly GPT-class models. We live in world where both products and services might become obsolete overnight, and we should also remember that many countries deliver projects successfully without the involvement of a Quantity Surveyor.

Before we get into this debate, it is worth summarising the reason Quantity Surveyors are required. Possibly, although it’s up for discussion, the primary value a Quantity Surveyor adds is our ability to provide assurance and advice through our valuation of construction work. We sit neatly in the triangle of engineering, financial and legal, and we are able to piece things together to build a picture of a project that clients and employers value and trust.

However, whether we want to see it or not, there is an existential threat facing us and many other professions. Is there a world where a client goes to an AI tool, describes their concerns and requirements, and obtains a reliable and credible valuation? For some that may be an impossible contemplation but run the test yourself and see if you think the answer is accurate. The reality is this is no longer an impossible contemplation because the technology is not theoretical anymore and the processing power and accuracy of the newest generation of models, for example, is incredible.

In the face of this reality, all is not lost and there is almost certainly hope and opportunity. We have become, whether we decided to be or not, the ones who hold together poorly defined project processes and vast amounts of polluted and unstructured data. Can we liberate ourselves from the administrative sludge we find ourselves in and towards value-add advice with real-time valuations that allow fast and effective optioneering during the design process? Can we compliantly and diligently administer multiple contracts with reduced resources, and invest more of our time in risk identification and management?

That is the challenge ahead of us and the one that will be explored in this article.

The project assurance battleground

For a long time, the ability to provide project controls and project assurance services was protected by the effort it took to do them well. Complicated spreadsheets, unstructured data, undefined processes are the domain of the QS who will read thousands of pages, reconcile costs from clunky financial systems, and check drawings and models for changes that might be variations.

We are the kings and queens of information chaos and, whether they would admit or not, most Quantity Surveyors love it! Our willingness to do that work where most just wanted the output, and our unique collection of skills, meant clients and their teams look to Quantity Surveyors to assure them that projects were on track or provided guidance where they had gone, or were at the risk of going, off track.

With the development of generative GPT-based tools, we have a very strong and able competitor. Documents can be read at speed, and the user can ask specific questions relevant to them and retrieve answers, financial data can be reconciled in seconds with only a few prompts, and large volumes of design information can be scoured and checked. And with that, the advice we give becomes more accessible.

So where is the competitive difference? Why would a client or employer cast aside their Quantity Surveyors and embrace the technology without caution?

The answer lies in reliability, credibility and a willingness to stand behind the answer given. The GPT will always carry a disclaimer, and it is unlikely a negligence claim would stick if made against the GPT. If you are a client making a multi-million-pound decision or involved in a high-stakes decision then you are looking for advice that de-risks your decision whilst giving you something, or someone, to fall back on if things don’t work out as planned.

The value enhancement of the Quantity Surveyor’s service, through increased productivity and increased output quality, will come in the ability to provide project assurance advice and services with the use of GPT. We need to demonstrate that we can use these tools responsibly and that we can mitigate the risks associated with their use to give clients confidence that they need a professional user of GPTs to obtain advice rather than go straight to a GPT for the advice.

Where GPT competes and where QSs must lead

It may not be that GPT trumps a Quantity Surveyor with its advice, for the reasons stated. It is more likely that other construction professionals use GPT to provide project assurance services. For example, it is not inconceivable that a project manager on an NEC contract could use a GPT to assess a compensation event.

If you want to stay relevant as the world evolves, then there are four project assurance areas that, if a QS can master, will ensure their skills are fit to be applied in the current and future workplace.

First are standards. We know what good looks like and have the ability to unpick a valuation with a few simple questions. We know what data is useful, what data is less than useful. And we know what the process that should be followed to achieve the outcome. The Quantity Surveyor’s nose for something that just doesn’t look or feel right will always be an asset.

Second is provenance and traceability. With a GPT chat, you will know the instructions that were given and the conclusion that was reached. However, if we think about that in the context of a report, would you be happy if that was all you received and, if you wanted to, you didn’t have the opportunity to read the methodology, the analysis and the findings? A QS should be able to work with a GPT to improve the quality and efficiency of their output without denying a stakeholder the opportunity to understand the process.

Third is judgement at the point of risk. The judgements we make are often based on experience and tacit knowledge. Sometimes you can’t explain why you know something will or won’t happen, you just know. This is valuable because you are able to provide options or scenarios and work with your client to understand their relevance and the solution. We might use GPT as a testing ground to widen the lens, but that human interaction will always be valued.

Fourth is disclosure. If the advice turns out be wrong and there is an argument of negligence, the GPT owner will not be underwriting their advice. That said, Quantity Surveyors may need to be open about where they have used GPT to demonstrate that it was responsible and that they have complied with professional standards. A balance needs to be struck here because the extent of disclosure could be proportionate and relevant.

We need to embrace our project assurance skills and combine this with the processing power of AI to deliver even better outcomes for clients. If we do this, the technology becomes a lever for quality as well as speed.

How to train your dragon

It might seem strange to think about a GPT as a dragon but there are some similarities. A well-trained dragon can be incredibly effective whereas an untrained dragon could cause untold destruction.

To train the GPT, you need to think through the journey from pilot to integration because it is fraught with challenges. However, it is a journey that is worth taking and, sooner or later, is one that you will have to take if you want your skills to stay relevant in the market.

When you look at your workload, or your team’s workload, what frustrates you the most? Where do you look at something and say the time required to perform that task is taking too long and represents the biggest opportunity to reduce waste. Then when you think about the problem, is it solvable through the intelligent use of technology?

Through asking these questions, that is where you will start to work out where the best returns can be achieved and, through a quick feasibility review, you can conclude whether the effort is worth the reward. Remember you don’t have to start big and you can start small to learn and get some wins, before progressing to some of those more challenging issues.

The solution you desire will influence your decision as to the information that your GPT relies upon. You could use information available on the website, but it is no secret that the internet is full of all kinds of weird and wonderful information with people giving their opinion, sometimes quite extreme opinion, on just about everything. So how do you know that the information it is using is accurate and can be relied upon? You could limit the available information to a smaller pool of documents but that could also have an adverse impact on the GPTs answers.

The important, but often overlooked part of the process, is testing. Too often, you can jump straight into use and sometimes widespread roll out. This can be dangerous with a GPT and could expose you to risks that you do not fully understand. So, test, test and test again. Make a record of the tests, documenting what you wanted to achieve versus what you actually achieved and what needs to be done differently next time. And once that is done, don’t forget to feedback to the GPT so that it can learn and improve.

At some stage within this process, you need to define the rules. The rules that users will follow and the rules that the GPT will follow. Some of this will be use guidance, but most of it will be governance type rules. It is inevitable that you will start with a skeleton based on information that exists (why not ask the GPT itself for guidance?) and this will develop over time through the testing phase and even once the GPT is being used.

The result from this should be a tool where you combine the unique skills of a Quantity Surveyor with the GPT power.

The market will decide our fate

It doesn’t feel like we are on the verge of a Kodak or Nokia moment. The role of a Quantity Surveyor continues to be in demand and that is likely to continue in the short to medium term. That certainty decreases over time because it is difficult to fully predict the impact that GPT tools will have.

If you are to flow with this change and remain relevant then embrace the technology, set the standards and demonstrate how you are using and controlling GPT to the benefit of your productivity and, ultimately, the service you provide to clients.

Whatever the future holds, good people using modern methods will never be short of demand for their skills.

Back to articles