AI’s Role in Workplace Strategy: Insights & Opportunities

In the rapidly evolving landscape of modern workspaces, technology plays a pivotal role in shaping the future of workplace strategy and corporate real estate. We had the opportunity to sit down with spaciv’s AI expert James Vaughan. Join us as we talk about the profound impact of technology on optimising workplace performance as we unravel the possibilities and potential that lie ahead.

Meeting scene with data network
Can you tell us about your background and current PhD topic? What specific areas do you focus on during your studies?

I’m currently working on a PhD on “Automated Reasoning” at the University of Edinburgh. I have been developing a theorem recommender system for automated theorem provers to allow them to solve more complicated problems.

In short, automated theorem provers are computer programs, designed to automatically find proofs for mathematical theorems without human intervention.

It turns out you can recommend theorems to proofs, much like how Amazon recommends products to customers. While Amazon uses past purchase history and the purchases of similar customers, we can leverage the knowledge about what theorems were used in similar proofs in the past.

What role does AI play in transforming workplace strategy and corporate real estate? Today and in the future?

At spaciv we’ve been using AI to supplement user contributions concerning their workplace needs – such as how much time they spend on individual work vs group work, or which days they prefer to work from home. We can estimate these needs from job roles, organisational structure, and the location associated with a position.

In the future, I see a similar approach working for the other side of the problem. Using machine learning to estimate how different space modules meet user needs from various attributes.

Can you discuss any of your recent insights demonstrating AI’s potential impact on workplace strategy and corporate real estate?

Perhaps unsurprisingly, we’re finding that there is an enormous amount of variability in how people work. That does put a limit on how accurately you can predict the needs of an individual position. But on aggregate, such as when predicting the needs of an entire team, that variance cancels itself out.

However, this trick only works if you don’t make any systematic errors in your predictions. Say, for example, your sample of the Sales department disproportionately worked from home, your model could easily overlearn that lesson, and in turn, changes to the office space may lead to Sales feeling cramped. But understating that evidence, assuming it is disproportionate, could lead to a similar systematic error and wasted office space.

One application of AI is to address this problem and balance your prior beliefs about how Sales departments work in general, against the evidence you have for how this specific one works. Furthermore, you can tease out any factors in your sample that could account for the discrepancy so that you can learn the right lessons from your data.

In your experience, what challenges or limitations do you anticipate when implementing AI solutions in the context of corporate real estate and workplace strategy?

Ultimately, I expect there’s going to be a limited amount of low-hanging fruit when it comes to explaining the variance in the data. There’s only so much relevant data that’s practical to collect from a client.

You also have to deal with change over time. The perfect office space for your needs next week may not work in 6 months if you need to take on new hires. Or your workload changes, and you find yourself in frequent brainstorming sessions in packed meeting rooms. Solving these problems without just throwing data at it is going to drastically drive-up model complexity.

What potential opportunities or advancements excite you the most leveraging AI for spaciv?

I think the big opportunity for spaciv comes with economies of scale. By taking on responsibility for many workplace strategies, we have the resources to bring to bear that a single organisation couldn’t muster on its own. Obviously, that would include even more extensive datasets. However, by combining these datasets with domain expertise, we can develop bespoke models that are able to continually increase workplace performance.

Ready to start?