By his own admission, Sebastian Deckker FRICS, is old school. Nearing 60, the Savills director still writes reports into a Word document, long hand, and rarely uses AI – which works for his specific job as a central London valuer, doing single units or small blocks of flats.
But despite his ‘traditional’ methods, Deckker recognises the phenomenal impact AI technology has had on his profession – particularly how it has fundamentally changed the way surveyors do their job. Especially with tasks involving location, general methodologies and calculations: “Where AI comes to the fore is with our viability team, who might be doing bigger schemes that are mixed-use with affordable housing, commercial elements and a gym.”
Today, surveyors blend their expertise with AI in day-to-day valuations, site monitoring and project management. The rapidly advancing technology can easily automate time-consuming and routine administrative tasks like lease abstraction, quantity take-offs and report drafting.
Using AI for certain tasks can slash data collection and processing time. This time saved allows surveyors to keep people at the heart of the process, making sure quality and client trust come first. “What's the most important factor in our work? It's probably trust, isn't it,” says Deckker.
At Arcadis, cementing trust is equally crucial, according to David Cullen MRICS, the company’s commercial director. He says the team that produces defined cost audits addresses client concerns around AI through “robust governance and contractual controls”. This includes entering into NDAs [non-disclosure agreements] that restrict and manage the use of sensitive information.
Clients are also exploring the use of AI themselves, which can be highly compatible with the work of Arcadis Cost and Commercial Assurance. This is particularly relevant in defined cost auditing, where large volumes of structured and unstructured data are processed and analysed. “We have the capability and knowledge to use these technologies responsibly,” says Cullen.
Overall, AI is used as a tool by the Arcadis team to support the review of documents such as disparate timesheet data, resulting in a 75% reduction in processing time compared to past manual efforts.
“This means we can work on building relationships with clients and their supply chains throughout the audit process,” says Cullen. “If we're recommending adjustments to costs and other issues, engaging with the client’s supply chain in a collaborative way and transparently explaining that information, the human factor is really where it comes into play and is such an [important] area.”
The need for people interaction and that professional scepticism cannot be replaced, says Cullen, as it is “paramount for defined cost audits.”
Who is accountable if AI gets it wrong?
An important consideration for any employer integrating AI into their work processes is who is accountable if AI gets something wrong. Does the buck stop with the individual or the company?
“The short answer is that accountability cannot be delegated to an algorithm,” says Claire Hookham, associate professor at the University of the Built Environment.
“From both a professional and legal perspective, responsibility ultimately remains with the qualified professional and the organisation that deploys the technology. If an AI-assisted valuation, cost audit or risk assessment proves to be materially inaccurate, it is unlikely that courts, regulators or clients will accept 'the AI made the mistake' as a sufficient defence.
“Accountability remains fundamentally human, even when the work is technologically augmented.”
Which means that companies must do all they can to make sure AI is being used properly and responsibly. Beyond RICS’ Responsible use of artificial intelligence in surveying practice, Hookham advocates “a broader organisational approach that integrates governance, ethics, professional competence and transparency”.
She adds: “Many organisations currently focus on what AI can do, rather than whether it should be used in a particular context. Responsible AI adoption requires asking both questions simultaneously.”
Among the practical approaches that companies using AI can implement are establishing formal AI governance frameworks, aligning with recognised international standards, and conducting AI impact assessments.
And when it comes to the wider issue of what is termed ‘AI slop’ and how it might affect the built environment, Hookham says it is already creating problems.
“While the phrase 'AI slop' emerged in relation to low-quality, mass-produced creative content, the underlying issue is much broader,” she says. “It refers to the rapid generation of outputs that appear plausible but lack accuracy, originality, critical thought or professional scrutiny."
According to Hookham, AI slop could manifest within the built environment in several ways:
- Automatically generated valuation reports that contain subtle inaccuracies
- Generic feasibility studies lacking contextual understanding
- Poor-quality project documentation
- Risk assessments that overlook site-specific factors
- Building performance analysis generated from incomplete or biased data
- Procurement and contract documents containing unverified information
“The danger is that such outputs often appear convincing,” she says. “AI is exceptionally good at producing information that sounds authoritative, even when it is incorrect. In professional practice, this can create significant risks because errors may not be immediately obvious.”
“What's the most important factor in our work? It's probably trust, isn't it” Sebastian Deckker FRICS, Savills
Turbocharging work
While acknowledging work can be “turbocharged” by using AI, as it increases the amount of data being reviewed, and the amount of compliance issues identified as a result, Cullen says: “The delivery of the message by a person is important because there are relationships that must be managed.”
Similar AI tools are increasingly being employed across the sector, so the tech has become a leveller, says Chris de Gruben FRICS, senior director and head of AI in property at Artefact. He was part of the expert working group on RICS’ Responsible use of artificial intelligence in surveying practice, that came into effect in March.
For instance, CBRE uses Ellis AI; an AI co-pilot that it says “automates routine tasks, proofreads, and analyses documents, and summarises articles and minutes to extract key findings [and ensures] prompts and documents remain secure. [This] saves employees vast amounts of time and allows them to focus on more value-added tasks”.
JLL, meanwhile has JLL Falcon, a proprietary purpose-built AI platform for commercial real estate that the company claims“ will leverage the best of human and machine learning, generating insights for employees and clients”.
As a result, what now distinguishes one company from another lies with its people, says de Gruben: “It’s what makes you competitive, what makes you good.”
What AI does, says de Gruben, is enable employees and businesses to focus on the human element – the relationships, the networking, and the expertise people bring to interpreting AI outputs into something understandable.
Embracing the change that comes from AI technology is now part of how the workplace evolves, as Deckker observes: “When I first started there wasn't even email, now I’m streaming an entire TV series on my iPhone. But there will always be the need for the human factor, and there's no substitute for picking up the phone or meeting for coffee or lunch.”
Hookham adds: “The future value of built environment professionals will not lie in competing with AI's ability to generate content. Rather, it will lie in their ability to assess quality, challenge assumptions, exercise professional judgement, and make accountable decisions in increasingly complex environments.”
No matter what the task at hand for chartered surveyors, the best use of AI is not to take humans out of the equation but to keep them firmly at the centre of it.
Responsible use of artificial intelligence in surveying practice
A new RICS professional standard, effective 9 March 2026
