AI is one of the fastest growing technologies and widely tipped to start a fourth industrial revolution. Last time out I explored how AI is changing how organisations make decisions. In this blog, I explore how AI can be applied into a risk assessment process, augmenting and complimenting the skills of existing professionals. 

To begin let’s breakdown the risk assessment process into a 4 stage framework: 

4 stage risk assessment process

1. Risk Recognition

Spotting risk is the foundation of all risk assessment. We humans are very good at spotting things out of the ordinary. If a supplier pays late or a key employee calls in sick we notice. The reason is down to biology. Our brains form neural pathways to lighten the cognitive load when repeating a task, freeing up brain power to ensure we’re more alert in our environment. In a physical environment (eg. assessing risk on a building site), this serves us very well. However, we don’t have the same evolutionary ‘training’ when it comes to digital environments (eg. assessing risk across a building supply chain)


The increase in workplace digitisation has created increasingly complex projects and organisations. The volume of data we can access has surpassed our natural computing power. Multiple stakeholders, gigabytes of data and numerous lines of communication, mean that without technology, risk recognition is very difficult. Before AI, technology has supported risk recognition by business function / department (eg. cost, health & safety, compliance). The benefit of AI is that it does not need to silo data to identify risk, and can identify patterns / irregularities in very large datasets quite easily. Given visibility of any multi-faceted process or project, AI has the potential to identify and flag risks that would otherwise go unnoticed.


2. Risk Assessment

Once recognised, a risk needs to be assessed. There are 2 areas that AI supports in the risk assessment stage. First, AI can surface additional information to the risk manager to inform the assessment. It is estimated that only 10% of business data is held in a structured format (spreadsheets and databases). AI has the ability to collate and surface information from the 90% of unstructured data (such as unseen reports and emails) which can illuminate the situation further. 

Second, AI is able to help mitigate natural human bias. Even highly skilled professionals are not immune to bias, and each risk manager will have a natural appetite for risk. This falls on a scale somewhere between being highly risk averse and being a risk taker. The other variable is that different projects require different tolerances to risk. Risk can vary by sector (public vs private sector) and project (house building vs plant pot manufacturing). The challenge to any large organisation is to be able to consistently assess risk when managers, sectors, and projects each come with a unique set of variables and tolerances. AI, given oversight of all risk managers and projects, can analyse trends and surface outliers where a risk tolerance diverges from the organisation, department or team standard. 


3. Risk Prediction

What is the impact and likelihood of a risk? In high-risk situations, humans typically err on the side of caution, but this may have unforeseen consequences. Research with the American College of Surgeons found that AI was better able to triage patients into postoperative intensive care than experienced surgeons. For a patient, staying out of the ICU when able to will reduce the risk of infection, and for the hospital, one patient less in the ICU helps fight issues like drug resistant superbugs. 

The scalability of AI is another key factor. AI is benchmarked against subject matter experts when it comes to its predictive capabilities (rightly so), but little attention is given to how unscalable subject matter experts are. The head of surgery can work in a handful of hospitals at best to improve postoperative ICU triage, providing the hospitals are geographically close, but trained AI can be deployed into every hospital in the world. The potential to improve risk prediction across whole industries is huge.


4. Risk Solution

In an increasingly complex and connected world, risk managers can find that traditional solutions are becoming less effective.  And new risk are appearing all the time. A challenge in coming up with new solutions to risks is that a lot of useful information gets siloed by departments. As explored in the risk recognition stage, the ability to illuminate more of a situation enables a manager to better predict risks, and better formulate the best solution. 


The potential of AI to analyse unstructured data is significant. This means risk managers can benefit from trend analysis and provide predictive modelling. If you’d like to explore how AI can support your risk assessment processes get in touch. We can arrange a free discovery session. so contact us at or call 0121 638 0760.