Artificial Intelligence in Real Estate Investment – AI versus Human Intervention

AI isn’t shy about joining real asset investing. It’s already active in other sectors, so it’s only a matter of time before it becomes more common here too. The question is: how much AI do we want, and how soon? Right now, there’s a tug-of-war between AI-driven analysis and human judgement. If investment decisions are being made on a pyramid of gap-filled, AI-generated assumptions, how safe are they?  

Definitions of AI and Applications in Real Estate

Before we discuss further, let’s start with clarifying some terms. ‘Artificial Intelligence’ is an umbrella term that covers any computer system that can mimic human intelligence in some way. Under this umbrella, you have Machine Learning, Deep Learning, and Generative AI, each serving different purposes, but all connected. 

Machine Learning (ML) is a way for computers to learn from data without being directly programmed. Instead of telling a computer every single rule, we give it examples and let it figure out patterns. For example, an ML model might look at thousands of real estate transactions and learn how to predict property prices based on factors like location, size, and market trends. ML models work best with structured datasets – clearly labelled and organised data like spreadsheets of energy usage or historical rental prices.  

Machine Learning is commonly used in investment analysis, risk modelling, and predictive maintenance. It has been used in real estate since at least the late 1990s, though early methods were more like traditional statistics. More advanced ML techniques became common in the 2010s as computers got more powerful and data became easier to access. 

Deep Learning (DL) is a specialised type of Machine Learning. It uses complex structures called neural networks, which are loosely inspired by the human brain. These networks have many layers (hence “deep”), allowing them to process massive amounts of data and detect patterns humans might miss. 

The main difference between ML and DL is that DL models learn by examples. That way it can automatically detect relationships in data similar to its training data without needing any human-designed rules. However, they also become less interpretable – we see what goes in and what comes out, but understanding exactly how the model reaches its conclusions is more difficult. 

For real asset investment, Deep Learning can be used to analyse complex datasets, such as historical asset performance across multiple markets, or to detect risk signals in environmental and operational data.

Generative AI (GenAI) is a subset of Deep Learning. Instead of just recognising patterns, it can create new things – text, images, music, or even predictive models. Tools like ChatGPT (which writes text) or DALL·E (which generates images) are examples of Generative AI. In real estate, Generative AI could help generate example investment scenarios based on past trends, generate synthetic data to fill gaps in historical records, or even create floor plan suggestions based on best practices. 
 

Artificial Intelligence, in broad terms, is not new for real estate, and the industry isn’t starting from scratch. The PropTech sector has paved the way for AI applications, and tools for investment management, building operations, portfolio oversight, and more has been around for a long time (like our software SIERA). That means the road is already laid for AI to jump in – if we know how and when to apply it.  

Risks and Possibilities with Generative AI in Real Estate

Generative AI is still early-stage, but it might soon handle tenant queries or summarise complex property documents. Additional opportunities could arise from the application of AI to underperforming assets, particularly where they do not already have sophisticated energy management systems. While we caution the blanket reliance on AI assumptions for real assets, not least as humans have always underestimated the impact of risk, we still see meaningful opportunities for rapid assessments as part of due diligence and action planning.  

There is also a potential security risk linked to the dissemination of strategic and highly confidential data sets outside of the original organisation; a risk that could grow exponentially if that data is then combined into an integrated dataset. 

Once confidential data leaves its original team, the risks of spying or theft jump. Nobody wants their strategic details landing in the wrong hands, resulting in steep financial or reputational fallout. 

Yes, caution is wise, but so is recognising the potential. 

How much Artificial Intelligence versus Human Intervention is needed for real assets?

It’s not simple maths: the models might be, but the underlying assumptions and exclusions are very difficult to set and do need Human Intervention, in our opinion. It feels like there is still a long way to go before we can trust the outcome of AI. In most cases we feel the provenance of data and quality controls are not yet part of a robust-enough approach. However, increasing public disclosure requirements will enable AI engines to link new trends and analyse the relationships between sustainability risks and values. AI is already being applied to investment analysis in other markets, so while we are lagging behind as usual, it will surely come – and quickly.  

So how do we move beyond “wait and see” and lock in a plan? If we think long-term, sustainability data should stand side by side with balance sheets. That means linking climate and investment metrics – while ensuring the numbers are good enough for AI to crunch. Tech dependence is a risk, but so is missing the boat on its benefits. The trick is pairing AI’s speed with solid controls on assumptions, data origin, and security. It’s time to ask: what steps protect value, cut risks, and keep operations afloat? For most, it starts with a strong data strategy. When the basics are in place – reliable collection methods, clear standards, and proper safeguards – AI can step in to help. Real assets are bound to see more of these digital tools. Done right, they could steer investors toward quick, smart decisions. Done wrong, they could spark big headaches. 

AI won’t replace solid human judgment, especially in a market that’s traditionally slow to trust big leaps. But if leaders treat data security, human oversight, and methodical adoption seriously, AI could be a real asset in shaping future investment decisions.