By Raj Singh, CEO
1. AI will change the demand for space in certain sectors and regions
2. Availability of data will (continue) to hinder adoption of predictive AI within real estate
3. Predictive AI will not be able to forecast trends in rents or asset prices
4. Generative AI will enhance, not disrupt, the roles of real estate investment professionals
5. Demand for real estate professionals will increase
6. AI’s impact within real estate will be much slower than in other industries
Recent rapid advances in artificial intelligence, particularly generative AI, have prompted many people in our field to wonder how these powerful new tools will impact the structure of the real estate capital markets and the daily activities of those working in the industry. As a technology provider dedicated to elevating the work of real estate professionals, these are questions that we ponder a lot at Altrio.
There is no shortage of blogs echoing the assertion that AI will “change everything”. Many vendors in our space have rushed to make exaggerated claims about the application of AI within their solutions. We believe most real estate professionals are (justifiably) skeptical. Eighteen months into the revolution, the tangible effects have been underwhelming.
In this paper, we will attempt to go beyond the hype and make some specific predictions about how the recent breakthroughs in AI will – or won’t – transform different aspects of the real estate capital markets.
1. AI will change the demand for space in certain sectors and regions
First, and most obviously, we expect shifts in demand for space across sectors and regions. From a sector perspective, as the demand for AI services increases and some jobs become less human intensive, the world will need more server racks, more power generation and distribution and fewer desks and meeting rooms. This will create opportunities for investment in data centers and infrastructure while adding to the pressure on the office sector.
Furthermore, regions with high levels of employment in industries that are either boltered or disrupted by AI will see rents buffeted across sectors as the general level of economic activity in those regions is positively or negatively impacted.
The timing of space market changes will likely not correspond with the investment cycle. As we have seen in previous “hype cycles”, asset prices in the affected sectors will probably overshoot economic fundamentals, resulting in an initial rally (or slide), followed by a correction, followed by a more gradual growth or decline in prices over the long term.
.png)
When considering how AI might change the way real estate investors, lenders and others in the market work day-to-day, it is necessary to disambiguate the term AI. When people use the term AI today, they often conflate very different technologies with different trajectories and potential impacts. We will start with “predictive” AI.
Predictive systems have existed for a long time and have been steadily improving over the past few decades, along with growth in computational power and the availability of data (within real estate the availability of data, not computing power, has long been the limiting factor when it comes to the adoption of predictive systems).
Predictive systems are essentially statistical models and “data science” is merely an evolution of statistics that harnesses new technologies and practices developed to cope with very large data sets.
One application of predictive AI within real estate comes from the ability of these algorithms to find patterns in large data sets. This can be used to identify markets that look like other markets that are deemed attractive or have performed well. Pattern matching can also be used to improve qualitative analysis by, for example, suggesting potential risks by matching investment opportunities to others that have historically been impacted by known issues.
2. Availability of data will (continue) to hinder adoption of predictive AI within real estate
While recognizing the potential of these applications, it is important to note that recent advances in AI have not meaningfully improved the efficacy of predictive analysis within the realm of real estate investing and are therefore unlikely to influence adoption. Rather, adoption of these technologies and approaches has been limited primarily by the lack of accurate, timely and consistent (“apples to apples”) information to feed the models.
Because real estate transactions are completed via the exchange of unstructured documents (typically over email, if not in person) rather than anything similar to the electronic platforms commonly used in other sectors, it is impossible to obtain complete, consistent and accurate information on these transactions. Without this data, the application of predictive AI within real estate capital markets remains impractical.
Some have made the argument that real estate professionals have been slow to adopt predictive models because they simply do not appreciate their potential. However, we don’t believe this. If there were examples of the technology making a clear and material difference to the performance of participants in the market, the natural instincts (and economic imperative) to compete would compel all firms to pay attention and follow suit.
3. Predictive AI will not be able to forecast trends in rents or asset prices
One of the most commonly cited potential applications of predictive AI within real estate is the forecasting of changes in rents and asset prices. We do not believe this is currently feasible.
Predictive AI models produce outputs from inputs based on having been trained on large sets of similar inputs and outputs. Typically, these models generate a single output, which is always wrong: when an algorithm generates a predicted change in rents, valuations or other metrics, the only thing the recipient of this prediction knows for sure is that the reality will be different.
As an investor considering a potential investment, it’s more valuable to have a range of probable outcomes along with a probability distribution. However, the real estate capital markets – at the intersection of space markets (influenced by myriad economic factors) and the markets for both debt and equity – is complex. Any model seeking to generate multiple scenarios must contend with the complex correlations between a multitude of inputs. As already stated, the data needed to accurately model these cross correlations is not available, making it difficult to create accurate probabilistic predictive models.
Generative AI, which we will discuss next, is not designed to do the type of analysis discussed above. As the name implies, generative AI is designed to create not analyze and as such has a very different set of potential applications within real estate investing.
4. Generative AI will enhance, not disrupt, the roles of real estate investment professionals
.png)
Generative AI and large language models have improved by leaps and bounds over the past eighteen months and have demonstrated an ability to significantly enhance many tasks that were previously considered the exclusive realm of human experts.
Fields like law, medicine and education – all of which, at their core, are based on digesting, synthesizing, structuring and presenting large sets of information look likely to be severely disrupted.
How does this relate to the world or real estate investing? One could argue that assessing a real estate investment is similar to assessing a contract or diagnosing a patient’s condition. Certainly there are aspects that are similar: for example, understanding a real estate asset’s potential value does involve analyzing a lot of information about the asset, the tenants, operating costs and local market context, etc.
However, the job of a real estate investor differs in many important ways. Investors operate in a dynamic environment in which the outcome of their decisions depend on the decisions of others outside their firms. Developments in real estate capital markets depend on a set of ever changing factors in both the market for physical space and the market for capital, not a set of relatively static facts as in the case of law and medicine.
A doctor determining a diagnosis, a lawyer reviewing a commercial agreement or a teacher preparing a lecture all operate in a context far less dynamic than the context in which a real estate investor operates.
This is not to say that generative AI will not be leveraged by real estate investors to increase their productivity. The ability to rapidly synthesize unstructured data and query that data using natural language will allow investors to churn through analysis much faster and spend more time considering qualitative and competitive dynamics.
5. Demand for real estate professionals will increase
Some have argued that the power of new LLMs will result in analysts being replaced by agents who can more quickly consume, digest, process and present information about prospective investments. We don’t agree. We believe that by increasing the productivity of real estate investors and lowering the barriers to entry (i.e. the effort/cost required to participate in the market), AI will lead to an increase in demand for real estate investment professionals.
Consider a simple example from outside the industry: a burger restaurant, where without the assistance of technology, a single line cook can produce 50 burgers per hour. If the installation of a new machine allows the same cook to produce 100 burgers per hour, the restaurant owner will need only half the number of cooks. So, it seems like the new technology will reduce the demand for grill cooks. But what happens next? The owner can now generate more profit per grill cook, which creates an incentive to hire more cooks and buy more machines – by opening a second restaurant. As the business grows and the number of locations increases, the total number of cooks needed to operate the machines across the chain of restaurants rises.
The dynamic illustrated in this example is well understood by economists. As the marginal productivity of a resource increases, so does demand for that resource. So, to the extent real estate investment analysts and associates begin to leverage AI to do certain parts of their jobs faster and better, they will become more productive and demand for their services (across society as a whole, if not within individual firms) will increase.
6. AI’s impact within real estate will be much slower than in other industries
The marginal productivity of human resources increases as more of their work is automated – up until their job is completely automated, at which point the total productivity very suddenly drops to zero. In some industries, generative AI will make it possible to completely automate jobs that were once performed by humans, resulting in mass unemployment in those sectors.
However, it will be incredibly difficult to automate the job of real estate investing because much of what occurs during the execution of a real estate transaction happens offline. Of course, investors, lenders, brokers, etc. use electronic forms of communication - emails, data rooms, etc. to complete transactions, but the lack of established standards and protocols, combined with the unstructured nature of most of the information involved in the transactions means that the capital markets are not “networked” in the same way many other markets are.
AI is fundamentally an application that must run on a platform. The power of an AI agent will always be limited by the capabilities of the platform on which it runs, specifically its ability to access real-time data and interact with others based on a common set of rules.
The offline, fragmented nature of the real estate capital markets and absence of accurate, timely data places significant limitations on the potential impact of AI agents within our industry – just as these limitations currently frustrate the efforts of the human agents who operate in the market.
Conclusion
At Altrio, we believe in the power of technology to elevate work and modernize industries. We also agree with the general sentiment that AI will have a large impact on many sectors of our economy and facets of human life.
However, we believe that within the real estate capital markets, data and connectivity are the two key factors that limit the productivity of people working in the industry and hinder their ability to benefit from advances in technologies like predictive and generative AI.
That’s why we remain committed to building tools that connect real estate professionals and bring processes online to generate normalized, real-time data that can power the application of current and future technologies.
We don’t believe that the work of real estate investors will ever be fully automated, but we see a future in which investors, lenders and other market participants will have the information and connectivity they need to assess and execute investments faster and more confidently than they can in the fragmented, offline marketplace of today.
.jpg)
