Anthony Scriffignano

The chief data scientist at Dun & Bradstreet speaks to Liz Lumley about data with a purpose, finding disruption in unknown things and fighting bias in AI. 

Google ‘Apple Watch saved my life’ and you will be met with a raft of stories about how the eponymous wearable computer detected an undiagnosed heart irregularity or dialled emergency services when its wearer experienced an accident or health incident. These features are made possible by collecting large amounts of very personal data about the user: blood pressure, heart rates and precise GPS locations, for example.

Career history: Anthony Scriffignano 

2002 Dun & Bradstreet, senior vice-president, chief data scientist

1998 Seton Hall University, adjunct professor

1996 Deloitte Consulting, senior manager

1987 Harrison Alloys, corporate controller/leader IT

While GPS location data is very helpful, potentially lifesaving information for an ambulance crew, it must be remembered that the same data could be very dangerous in the wrong hands — such as those of an abusive ex-partner, for example. 

The total amount of data expected to be created, captured, copied and consumed globally in 2022 is 97 zettabytes (one zettabyte is a trillion gigabytes) — a number projected to grow to 181 zettabytes by 2025, according to Statista. However, more data does not necessarily mean better analytics. How you collect and use that data should have “purpose”, according to Anthony Scriffignano, senior vice-president and chief data scientist at data analytics firm Dun & Bradstreet. 

Personalisation versus privacy

“We want our lives to be … surrounded by things that are increasingly customised, but then we also want privacy,” he says. “Those are sort of opposite things to want — it’s not a question of one or the other — it’s a question of being where you are in that continuum on purpose.”

Mr Scriffignano believes that the world is moving towards more transparency on what is collected and what is being done with our data. “It’s about the unintended consequence of being over-controlled,” he says. 

“If I had to permission every single use of every piece of data that I’m producing as I walked through my life, I would never get to do anything; I’d be constantly accepting or rejecting privileges.”

When looking at data usage in a business context, firms need to comply with an array of global laws and regulations, and these laws often do not make it easy for organisations to collect useful information on, for example, individual fraudsters. 

There’s a reason why banks are regulated. That’s where they keep the money, right?

“There’s a reason why banks are regulated,” Mr Scriffignano says. “That’s where they keep the money, right? It’s not necessarily a bad thing — it might create a tapestry that is more resilient to fraud; or more resilient to stopping the funding of human trafficking, drugs and things like that.”

However, one of the challenges for financial services when dealing with data is having to understand “the potentially conflicting regulatory requirements in different parts of the world”, he says. This job is “not something for the lighthearted. It’s a full-time job making sure that you’re compliant with things like that.” 

Getting regulation right

As an avid scuba diver, Mr Scriffignano has an analogy that he finds useful. “When you scuba dive, you have a tank of compressed air on your back and you have a regulator. That regulator keeps that highly compressed air from going into your lungs, at full pressure, which would kill you,” he says.

“So that regulator is pretty important. But if it cuts off all the air, then you can’t breathe — we’ve got to get the regulation right.”

As the pace and volume of data created and collected has increased exponentially, so has the use of advanced analytics and artificial intelligence (AI). While automation, rules-based algorithms and machine learning are all needed, if only to handle the sheer amount of data, Mr Scriffignano says he “cannot over-stress how important it is to understand the bias in AI methods”. A lot of this bias is due to many methods assuming that all data sets are true.

“All of our data is not true: people lie and data is incomplete,” he says. “If your model has synthetic data, you tend to build data that looks like the rest of the data — that makes it hard to find anomalies [and] causes a different type of bias,” he adds. 

While there are many different types of bias in AI, it is not really the AI that has the bias, adds Mr Scriffignano. “It’s the methodologies — the people that use these tools not necessarily thinking through what the preconditions are, what you have to believe in order to use a method like that.”

Dun & Bradstreet is all about using data to help their customers. Currently, Mr Scriffignano is looking at detecting disruption in places that you cannot see, such as closed or seemingly stagnant systems. An example of such a place would be supply chain disruptions — much like what happened during the Covid-19 pandemic, or when the Suez Canal was blocked by a stuck container ship. 

“We don’t finish getting disrupted by one thing before the next disruption happens,” he adds. “All of that causes perturbations in the accuracy, completeness and the timeliness of our data, so understanding how the data is changing with all of this change in the world is a very big job.”

PLEASE ENTER YOUR DETAILS TO WATCH THIS VIDEO

All fields are mandatory

The Banker is a service from the Financial Times. The Financial Times Ltd takes your privacy seriously.

Choose how you want us to contact you.

Invites and Offers from The Banker

Receive exclusive personalised event invitations, carefully curated offers and promotions from The Banker



For more information about how we use your data, please refer to our privacy and cookie policies.

Terms and conditions

Join our community

The Banker on Twitter