Technology
Causal AI – machine learning for the real world
Published by linker 5
Posted on November 6, 2020
4 min readLast updated: January 21, 2026
Global Banking and Finance Review
Company
Published by linker 5
Posted on November 6, 2020
4 min readLast updated: January 21, 2026
By Darko Matovski, CEO and co-founder, causaLens
Many businesses use machine learning algorithms to automate operational processes, solve complex data-rich business problems, and inform data-driven decisions. Traditionally, businesses in the financial services sector have depended on analysis generated by quants or machine learning platforms that rely solely upon historical correlations in order to make predictions about the economy. Such approaches have severe limitations, however, often doing little more than fitting curves to data with no understanding of the real world. These businesses are effectively driving forward by looking in the rear-view mirror.
This need no longer be the case, though. With an understanding of cause and effect, machine learning platforms can unlock significant additional value in data. A necessary step toward developing true artificial intelligence, the application of causal inference can enable businesses to access more precise predictions than ever before. And its benefits aren’t just limited to financial services.
Cause and effect in action
We tend to think of things in terms of cause and effect. If we can ascertain why something happened, we can change our behaviour to change the outcome of a similar situation in the future. When used as a tool, causal inference can enable AI algorithms to reason in a similar way, thereby addressing one of the biggest challenges around machine learning.
Machine learning is unable to adapt to environmental changes. In the case of predictions, they will severely overfit to the historical data. This is particularly true in the world of finance – the ultimate dynamic system.
Ideally, machine learning algorithms should be able to generalise, and adapt to new and previously unseen data. Enabling algorithms to understand causality improves that generalisation. Causal AI makes more accurate predictions in complex systems such as those found in financial services.
Unlike traditional machine learning approaches
In practice, a causal AI platform will retain the advantages of comprehensive automation – one of the key benefits of machine learning – allowing thousands of datasets to be cleaned, sorted and monitored simultaneously. Unlike traditional approaches, however, it combines this data with causal models and truly explainable insights – traditionally the sole province of domain experts. To further enhance its capabilities, intuitive interfaces can be used to harness unique human knowledge.
Darko Matovski
By way of illustration, Foreign Exchange (FX) technology provider CLS recently implemented our causal AI platform. By helping the company to understand relationships between its data and other datasets, it enabled the rapid identification of significant and unexpected changes in key factors associated with the FX markets during the COVID-19 pandemic. According to Masami Johnstone, head of information services at CLS, this was immensely valuable to its clients, enabling them to react quickly to market conditions, and enhance their investing strategies.
As well as this, many of our hedge fund clients have used the platform to evaluate datasets very rapidly and at scale. This helps them find profitable trading signals in hundreds of large datasets, leading to cost savings in data evaluation and increased profits from taking positions through listening to these signals. In the insurance industry, the technology can be used to predict claim developments, amount per claim, business defaults and macro-economic indicators leading to at least a 15 percent decrease in costs.
Causal AI represents significant benefits to other industries, too. By improving staff and bed allocation, for example, and predicting the spread of diseases in real-time, it could save healthcare providers up to 15 percent in operational costs. The oil and gas industry could enjoy savings of at least $200bn by optimising transportation and storage, and more accurately predicting supply and demand. And more than $500bn in food waste could be saved each year if we could better predict microclimate and demand.
A huge step forward
Causal AI is already being used to predict and optimise many businesses but, as the examples outlined above demonstrate, it has the potential to improve operations across a range of industries and, in time, could come to play a role across all of society.
Today’s world is changing faster than ever before. It’s vital that businesses are able to react quickly and adapt to unexpected circumstances and events – especially in times of political and economic uncertainty such as those we’re currently experiencing. But current state-of-the-art machine learning barely scratches the surface of what a business’s machines can do. A new category of machine learning, causal AI goes far beyond predictions, providing transparent causal insights and suggesting actions that can directly improve a business’s KPIs. By understanding – and acting upon – the importance of cause and effect, the advent of causal AI represents a huge step forward for businesses everywhere.
Explore more articles in the Technology category
