EuroFinance 2017 | Will Machine Learning And AI Take Away Treasury Jobs?

Corporate treasurers may not be out of a job yet, but they will have to adapt, or at least learn how to work with AI in their businesses

 In a Treasury Lab session at EuroFinance 2017, companies explained how  machine learning, AI and Big Data could change treasury operations in future.
In a Treasury Lab session at EuroFinance 2017, analytics companies explained how AI, Machine Learning and Big Data could change treasury operations in future.

Will corporate treasury be freed or terminated by artificial intelligence—machine learning, robotic process automation? Will it free up their time to focus on the more strategic and less mundane aspects of their business? Can they glean new intelligence from applying complex algorithms to data sets—identifying fraudulent transactions, reducing their accounts receivable?

Most discussions about artificial intelligence (AI) tend to be framed in the context of machines taking jobs away from humans. But at the opening panel discussion on the second day of EuroFinance in Barcelona, the debate turned more towards how AI, machine learning and Big Data can help humans do their jobs better and increase productivity.


Big Data is a term that is often bandied about. But what does it mean for corporate treasurers? With few corporations having launched projects to apply computational algorithms (machine learning) to large data sets to reveal patterns and trends, so far it seems to be the exclusive preserve of technology companies or fintech start-ups

But in a Treasury Lab session at EuroFinance on Wednesday in Barcelona  on Big Data, analytics companies explained how they could potentially help treasurers identify errors in bank charges or fraudulent transactions.

Today, Roxon says most incidences of fraud are identified by tipping hotlines. “Only 14% of fraud is caught by audits and software comes in at 3%,” he says. Using AI, companies like MindBridge hope to be able to compare more data points to identify fraud more quickly and easily.

 Another exciting application of AI to Big Data, which could have significant ramifications for corporate treasurers and their relationship with transaction banks, is bank fee data, which is already widely available in the US, and is starting to become available in Europe, Asia and Latin America. By applying analytics to these data sets, Bridget Meyer, senior director at Redbridge Analytics says they found more than 400 anomalies on the bank billing side, that were not detected by humans. The average dollar amount for these anomalies was $2,000.

“Banks are using this data to understand you as a corporate,” said Meyer, “Your treasury management system won’t tell you what you did well or what you did wrong, and what you are still receiving by fax.”

 But with data becoming a highly prized asset between corporates and their banks, questions are likely to arise as to who owns the data and how it is being used? The quality of the data is also important. As Roxon points out, AI is only as good as the underlying data. “The rule of garbage in, garbage out, still applies,” he says.

“Technology should free up people’s time,” says Adam Rutherford, writer, broadcaster and scientific advisor on AI and robotics. “Management has to recognize that to develop a healthy work environment.”

Whilst artificial intelligence has moved on from smart machines doing mundane, repetitive tasks, to more complex human-like behaviors such as beating humans at the strategy board game, Go, and algorithms that learn the more data you give them, instilling human-like qualities such as self-awareness and common sense into AI applications have not been solved yet. “[Self-awareness] is the last bastion of AI, which we may never conquer,” said George Zarkadakis, digital lead at risk management, insurance brokerage and advisory company Willis Towers Watson in the UK.

Corporate treasurers may not be out of a job yet, but they will have to adapt, or at least learn how to work with AI in their businesses. “Imagine a world,” says Zarkadakis, “where you can go beyond structured data and look at ‘dark data’—being able to predict risks that are not in your structured data and that can help you be a more creative person.”

The rise of the machines is likely to favour those individuals who are generalists, , says Zarkadakis, rather than specialists. “If you’re a generalist that can bring people together to collaborate and work with data, you are less likely to be terminated. If you want to survive in this new world, you have to understand Big Data and be a people person.”

The right collaboration culture is needed to make AI work in a corporate environment says Zarkadakis. But there is a dark side to AI. “Technology companies may have a different agenda to society [when it comes to using AI],” says Zarkadakis, “that may not necessarily be what society wants to do.”

For most corporate treasurers, the widespread application of AI within their own businesses still seems like a distant prospect. Most of the work in this area is still being done in labs set up by technology companies like Google and Apple.

One of the areas, however, where AI could have significant ramifications for corporations and their finance departments is in applying complex machine learning algorithms to data to  detect fraudulent transactions, for example, or more simply, getting the right information to the right person at the right time so they can do their job better.

Advances in computing power (Quantum computing) and data storage (in the Cloud) makes applying complex algorithms to large data sets to detect patterns unrecognizable to the human eye, more feasible than ever before.

However, if AI is to move from the lab to the real world and corporate treasuries, then a number of things need to happen, says Chris Wigley, a partner at McKinsey and COO of QuantumBlack, a McKinsey data analytics company. Firstly, he says companies need to identify data sets they can work with—they might be in their enterprise resource planning systems or they may have to create new data. Companies need to identify the “high value” opportunities, says Wigley, find the data sets that reflect reality, and build a portfolio of trials (not all of them will be successful). 

Data security is also key, says Wigley. QuantumBlack recently appointed a chief trust officer and he says they try and instil data security into the “muscle memory” of all their employees–“it’s like cybersecurity “Kung Fu.”