Anyone yearning for the quiet life should look away now, says Dr David Nicholson. Today’s horizon brims with geopolitical challenges and ongoing transformational change – not least the deployment of artificial intelligence.
Have you ever read David Halberstam’s 'The Best and the Brightest'?
It’s describes how American leaders, widely viewed as multi-talented and gifted, managed to steer the US into the disaster of the Vietnam War.
It has long been seen as essential reading for anyone interested in strategy and decision-making and I can’t help but think it should be on the reading list of our current crop of business and political leaders – both domestic and international.
After all, there’s not exactly a shortage of complex issues piling up in their respective in-trays. From urbanisation to climate change, changing demographics to globalisation, challenges loom large while solutions remain elusive.
And prominent among the changes sweeping across the political, economic and business landscapes is the ongoing technological revolution underpinned by rapid advances in artificial intelligence (AI) and machine learning.
Build a positive bully pulpit
It’s not surprising that AI has rapidly moved into the mainstream. After all, no sector is immune to its rise – it is all around us and is staying that way, for good or ill. This means that leaders – in government and business – have a vital role to play in fashioning a future where AI can raise standards, improve efficiencies and strengthen services. By understanding its importance and potential, and engaging widely with key stakeholders, they can help construct a framework whereby developers and designers can harness AI for the public good.
There are already countless examples on which to build. Last year, a House of Lords Select Committee on AI published a report which recommended that that the UK needs to take a leading role in AI ethics and that the technology ‘should be developed for the common good and benefit of humanity’.
Globally, too, there is evidence aplenty. Africa’s Wildtrack programme is using AI to match crowd sourced images of animals to enable wildlife experts monitor and track endangered species. In the US, the state of Maryland has deployed an AI traffic system that is projected to reduce commutes by 15 per cent. And Estonia – long a pioneer for all things digital – is now developing a legal framework around AI’s deployment.
The fight against financial crime
In my own field of fraud and financial crime, AI is undoubtedly a force for good. Already, it is helping spot suspicious transactions and behaviour by blending analytical firepower with the intuition and skills of human investigators.
But as the last few years have taught us, more advances are set to materialise as we look ahead into the rest of this year and beyond.
For example, 2018 witnessed growth of ‘Explainable AI’, which aims to peek inside black-box machine learning algorithms to explain their decisions. This is particularly pertinent when business users want to understand the reason behind a fraud alert or a regulator wants to interrogate a financial institution’s compliance approach. Explainable AI techniques – such as LIME (Local Interpretable Model-Agnostic Explanations) and SHAP (SHapley Additive ExPlanation) started to draw attention and flourish in 2018, and I’m excited to see if this journey progresses into their routine application in operational systems.
Another key theme that will continue through the remainder of the year is the democratisation of data science and the empowerment of citizen data scientists.
Citizen data scientists can now routinely run complex data analytics without having to worry about navigating through the diverse range of possible analytics and understanding their associated parameter meanings and configurations – an AI-enabled machine can do this for them. Now, they are poised to benefit from machine intelligence to also get data swiftly and accurately to the stage where analytics can mine it for information – a significant pain point that we should not expect a citizen data scientist to suffer.
The pain arises because real-world data in financial crime detection is neither homogenous or clean: it is a mixed bag of variable quality that has to be wrangled and mangled into shape. Achieving this manually, in a spreadsheet or with scripting, requires patience and expertise.
So I am excited to see if the success of automatic machine learning can be extended toward the data itself, to automatically perform low-level functions such as cleansing the data and extracting useful features for machine learning, thereby freeing citizen data scientists to focus on higher-level tasks.
From policy to impact
But what does all this activity mean for those in the corridors of power? Clearly, they don’t need to master the mechanics of AI technology but they do need to understand how to channel it and ensure it is not used for nefarious purposes.
Their task now is to ensure that this still nascent technology heralds a positive future and one not shaped by machines run amok.
About the author
Dr David Nicholson is Financial Crime Data Science Leader at BAE Systems Applied Intelligence.