Here Are The Practical Gains And Existential Questions That AI Brings To Real Estate

Car park usage. Confidentiality. Who gets sued when a valuation is wrong, and what is the risk of a meltdown when everyone piles into the same type of deal?

The increasing use of artificial intelligence to find solutions in real estate is set to have impacts prosaic and profound. And as the potentially transformative technology gathers momentum, the industry is only in the very early stages of working out the problems AI might solve and how it will do so.

“We're trying to start with the problem we’re trying to solve, as opposed to the latest technology that someone might be showcasing to us,” LGIM Real Assets Senior Strategist Matt Soffair told the audience at Bisnow London’s AI in the Digital Age of Real Estate event.

Speakers at the event talked through multiple ways in which their firms are finding practical and profitable uses for AI and machine learning technology. They also pondered some of the larger questions AI poses for a sector that has historically made confidentiality and asymmetry of information the norm. 

Many of the applications being used centre around the design and development process, panellists said.

“Design is definitely one area that I'm super excited about from an AI perspective,” Concrete Ventures partner Rajdeep Gahir said. “The detailed design and technical design process is generally not something that I have found architects enjoy too much. It also seems to be an area where you spend a lot of time and your margin ultimately gets compressed.”

Concrete, a venture capital investor, is working with companies that use AI to turn concept drawings created by an architect into detailed technical designs, saving time and money. 

Basil Demeroutis, managing partner of developer and investor Fore Partnership, cited several AI tools the company uses. One allows users to plug in the size and parameters of a development site and add information about the on-site planning constraints to determine what size and shape of building could be built there. 

Another tool takes plans from very early stage building information modelling and creates a fully designed mechanical and electrical system, taking existing designs and optimising them for a 20% savings in design costs.

LGIM’s Soffair said the company was generally using AI technology in the early phases of designing interiors of buildings that need to be refurbished, adding the increased speed makes a costly process cheaper and more efficient.

But going forward, AI tools could be used to comb through development appraisals to work out which schemes might fit a particular firm’s criteria, a process that is complex and time-consuming. 

Sustainability is another major possible beneficiary, panellists said.

Once upon a time, the only data point a building owner had to understand how much energy a building was using was a quarterly utility bill, Demeroutis said. Now, with hundreds of sensors around a building providing real-time data, millions of data points a year are generated. AI can sift that data for patterns and provide information to reduce that usage, he said. 

LGIM is using AI tools to scrape data from sources like social media and review sites to find leads for its coworking business, a process that has the potential to disrupt the livelihood of brokers in the sector, Soffair said. Another tool utilises satellite imagery of car parks at its out-of-town retail parks to work out how they could be used more efficiently, including the potential for adding income-producing property to the sites. 

In the more abstract world of leases and contracts, AI has the potential to extract key information from documents that might be hundreds of pages long.

“People love checking the work of others,” Stonal CEO Robin Rivaton said. “If, for instance, you need an exact list of all the equipment in your building or the key information in the leases in your portfolio, that might mean reading through thousands of pages of documents. But AI can create tools that allow you read through a summary and say, ‘Yes, this is useful. No, that isn’t.’ And it decreases the time needed to do the job.”

The same is true of reading and writing requests for proposals, he said.

In terms of who within an organisation will need to be able to use AI tools, particularly generative AI like ChatGPT, panelists broadly agreed that the utility of the technology lends itself to some roles more than others.

But issues around confidentiality mean that organisations will likely have to develop their own large language models, the AI process that powers tools like ChatGPT, rather than relying on those created by companies like OpenAI or Google. The reason? Putting confidential information into ChatGPT essentially makes it public, and that information could become part of answers to questions entered by other users. JLL and Cushman & Wakefield have created their own internal company systems as new models for CRE usage proliferate. 

Though the potential of AI is vast, its accelerating use raises broader philosophical points, panellists said. 

Valuation, or the process of comparing the prices paid for similar buildings, has long been cited as an area ripe for disruption by AI. But if a valuation is undertaken using an AI tool and that valuation ultimately leads to a lawsuit, who gets sued is an unknown, Pi Labs Head of Research Luke Graham said. 

There are also worries that if large numbers of companies use similar tools to asses development or investment opportunities, everyone will pile into the same part of the market.

That phenomenon is already apparent in the rush to get into top-performing sectors like logistics or life sciences, creating potential systemic risk, according to one questioner from the audience who expressed concern about concentrating risk in certain sectors or strategies.

Panellists were not convinced by the argument. Real estate is a heterogeneous sector in which asymmetry of information and differences of opinion make a market, they said. After all, not everyone can buy the same building. 

“I think this comes back to the point of valuation. What the building is worth still sits in the eye of the beholder,” Schroders Head of European Real Estate Research Oliver Kummerfeldt said. 

“We try to develop all these sophisticated models, but if you asked 20 forecasters, they would give you 20 different forecasts. What the future holds is still something where you need to have an opinion and a conviction.”