Advances in artificial intelligence are giving organisations in both the public and private sectors increasingly powerful forecasting capabilities. How much further down this predictive path is it possible for them to go?
Minority Report, Steven Spielberg’s 2002 sci-fi thriller based on a short story by Philip K. Dick, explores the concept of extremely proactive policing. The film, starring Tom Cruise, is set in 2054 Washington DC. The city’s pre-crime department, using visions provided by three clairvoyants, can accurately forecast where a premeditated homicide is about to happen. The team is then able to dash to the scene and collar the would-be murderer just before they strike.
While police forces are never likely to have crack teams of incredibly useful psychics at their disposal, artificial intelligence has advanced to such an extent in recent years that its powerful algorithms can crunch huge volumes of data to make startlingly accurate forecasts.
Could a Minority Report style of super-predictive governance ever become feasible in the public sector – or, indeed, in business? If so, what would the ethical implications of adopting such an approach be?
There is a growing list of narrow-scope cases in which predictive analytics has been used to fight crime and save lives. In Durham, North Carolina, for instance, the police reported a 39% fall in the number of violent offences recorded between 2007 and 2014 after using AI-based systems over that period to observe trends in criminal activities and identify hotspots where they could make more timely interventions.
AI has also been used to tackle human trafficking in the US, where it has helped the authorities to locate and rescue thousands of victims. Knowing that about 75% of child trafficking cases involve grooming on the internet, the government’s Defense Advanced Research Projects Agency monitors suspicious online ads, detects coded messages and finds connections between these and criminal gangs.
In Indonesia, the government has partnered with Qlue, a specialist in smart city technology, to predict when and where natural disasters are most likely to strike. Its systems analyse flood data collected from sensors and information reported by citizens. This enables it to identify the localities most at risk, which informs disaster management planning and enables swifter, more targeted responses.
While all these cases are positive examples of the power of predictive AI, it would be nigh-on impossible to roll out a Minority Report style of governance on a larger scale. That’s the view of Dr Laura Gilbert, chief analyst and director of data science at the Cabinet Office. “To recreate a precognitive world, you would need an incredibly advanced, highly deterministic model of human behaviour – using an AI digital-twin model, perhaps – with low levels of uncertainty being tolerable,” she says. “It’s not certain that this is even possible.”
An abundance of information is required to understand a person’s likely behaviour, such as their genetic make-up, upbringing, current circumstances and more. Moreover, achieving errorless results would require everyone to be continuously scrutinised.
“Doing this on a grand scale – by closely monitoring every facet of every life; accurately analysing and storing (or judiciously discarding) all the data collected; and creating all the technology enhancements to enable such a programme – would be a huge investment and also cost us opportunities to develop other types of positive intervention,” Gilbert says. “This is unlikely to be even close to acceptable, socially or politically, in the foreseeable future.”
Tom Cheesewright, a futurist, author and consultant, agrees. He doubts that such an undertaking would ever be considered worthwhile, even in 2054. “The cost to the wider public in terms of the loss of privacy would be too great,” Cheesewright argues, adding that, in any case, “techniques for bypassing surveillance are widely understood”.
Nonetheless, Vishal Marria, founder and CEO of enterprise intelligence company Quantexa, notes that the private sector, particularly the financial services industry, is making great use of AI in nipping crimes such as money-laundering in the bud. “HSBC has pioneered a new approach to countering financial crime on a global scale across billions of records,” he says. “Only by implementing contextual analytics technology could it identify the risk more accurately, remove it and enable a future-proof mitigation strategy.”
Alex Case, senior director in EMEA for US software company Pegasystems, believes that governments and their agencies can take much from the private sector’s advances. Case, who worked as a deputy director in the civil service from 2018 to 2021, says: “The levels of service being routinely provided by the best parts of the private sector can be replicated in government. In contrast with the dystopian future depicted in Minority Report, the increasing use of AI by governments may lead to a golden age of citizen-centric public service.”
Which other operations or business functions have the most to gain from advances in predictive analytics? Cheeswright believes that “the upstream supply chain is an obvious one in the current climate. If you can foresee shortages owing to pandemics, wars, economic failures and natural disasters, you could gain an enormous competitive advantage.”
The biggest barriers to wielding such forecasting power are a lack of high-quality data and a shortage of experts who can analyse the material and draw actionable insights from it. “Bad data can turn even a smooth deployment on the technology side into a disaster for a business,” notes Danny Sandwell, data strategist at Quest Software. “Data governance – underpinned by visibility into, and insights about, your data landscape – is the best way to ensure that you’re using the right material to inform your decisions. Effective governance helps organisations to understand what data they have, its fitness for use and how it should be applied.”
Sandwell adds that a well-managed data governance programme will create a “single version of the truth”, eliminating duplicate data and the confusion it can cause. Moreover, the most advanced organisations can build self-service platforms by establishing standards and investing in data literacy. “Data governance enables a system of best practice, expertise and collaboration – the hallmarks of an analytics-driven business,” he says.
Gilbert offers business leaders one final piece of advice in this area: recruit carefully. She argues that “a great data analyst is worth, at a conservative estimate, 20 average ones. They can often do things that any number of average analysts working together still can’t achieve. What’s more, a bad analyst will cost you both money and time.”
And, as Minority Report’s would-be criminals in discover to their cost, time is the one resource that’s impossible to claw back.
This article was first published in Raconteur’s Future of Data report in October 2022