Recently, Napier’s Head of Analytics, Dr Janet Bastiman, joined an expert panel at the two-day FinCrime World Forum – a GRC World Forums Experience event.
The panel were tasked with answering the highly topical question, is artificial intelligence the ultimate weapon against financial crime?
The outcome was that everyone agreed AI is not a silver bullet for fighting financial crime. Although deemed to be an important and powerful weapon, the consensus is that it is not the only one.
Here are the main themes that cropped up:
AI needs good data and effective rules
AI is almost as important as the data that feeds it. You need good quality data to get the most out of the system.
Also important are good quality, effective rules to be sure that all the known knowns are captured.
Artificial intelligence is a wonderful tool for fighting financial crime, but basic housekeeping needs to be done first.
Traceability, explainability and interpretability are essential for AI to do its job properly
“With great power comes great responsibility”.
For AI to be most effective, it is important that you understand the technology and demand explainability in any AI solution you choose.
A score on its own is not acceptable. You need to know and understand a decision taken by artificial intelligence, so you can decide what further action would be appropriate.
While for some use cases a black box may be acceptable, an opaque box allows you to see the insights alongside an explanation described plain language. This means any user can understand why an alert has been created and what is unusual.
This is important. If humans understand a decision made by artificial intelligence, they will be able to trust it.
Artificial intelligence is just one of many tools in the toolbox
Artificial intelligence is just one of many tools in the toolbox for fighting financial crime.
Trying to apply artificial intelligence across the whole challenge can be counterproductive; artificial intelligence doesn’t have to solve the whole problem. It can be a key part of a solution, applied in different ways in different places.
Artificial intelligence can reduce false positives but there are other use cases too
Artificial intelligence is most commonly used to reduce costly false positives in the financial crime. But this isn’t the only benefit.
Other use cases of artificial intelligence include analyst investigation automation, alert triaging and big data analytics to identify anomalies and crime. It is important to link human intelligence with machine intelligence to get the most out of any system.
More collaboration is needed to fight financial crime
Finally, there was strong consensus that wider industry collaboration between tech companies, banking and law enforcement is needed. Data sharing should be a priority to improve intelligence and detection.
It was suggested that data should be centralised with every bank using a single platform to detect financial crime. Such collaboration would ensure criminals can’t hide.
Lastly, the need for fundamental interconnectedness is important to clamp down on crime. Dr Janet Bastiman concluded the discussion by saying, “we need to spread information about criminal activity immediately, so avenues are cut off almost as quickly as they are created."
The expert panel comprised:
Be the first to know about Napier’s events