Forget fake news: why you need to know about fake data and how it can affect you

 

Fake data is increasingly becoming an issue for companies - here's ever ything you need to know.By now, we’re all very much aware of the issues surrounding fake news. In 2016, fake news stories claimed the Pope was endorsing Donald Trump for the US presidential election, or that a pedophilia ring was operating out of a Washington DC pizza restaurant.

But, what you might not have heard about is fake data. Just as much of a problem as fake news, fake data has the potential to affect companies and people around the world.

Here’s what you need to know about the phenomenon.

What is fake data?

According to a report by global consulting firm Accenture, fake data is data that is unverified, maliciously tampered with, or just plain wrong.

Across the world, 97 per cent of business decisions are made using data. But the issue is what happens when these decisions are made with the wrong data and what that could mean for people and consumers.

Is this a new problem?

The topic of fake data is not new, but it’s something that companies are increasingly becoming concerned about.

In the past few years, the improvements in artificial intelligence (AI) technology means more and more businesses are using autonomous decision making, which is when machines can make decisions on the basis of the data in front of them.

“Before there would be a human authorising the decisions but that is now very much streamlined through automation,” Accenture’s MD of technology, Yann Lepant, tells the Standard.

If a company wants to grow and compete on a global scale, they have to use data. AI and machine learning algorithms are used in this way to constantly make decisions that humans aren’t always on hand to do.

The problem is what happens when the data being used is wrong.

How can fake data impact me?

There are a couple of instances when fake data has had a real impact. For instance, a few years ago Amazon had an issue with fake reviews – when third-party sellers were paying people to leave false reviews on products.

“This became quite a problem as it was inflating artificial ratings on the platform and affecting Amazon’s own credibility,” says Lepant.

This, in turn, was affecting consumers; many people make decisions to buy a product based on reviews. But when a product review is fake, this can have a negative impact on Amazon shoppers as they are buying a product based on false information.

It extends into politics too. In the US, the state of Indiana began using an automated system to flag people who may be registered to vote in more than one state.

If there were two registered voters with the same name and birth date, the AI would automatically remove them from the voting register to prevent duplication.

However, 99 per cent of the alerts made for fraud were inaccurate, making mistakes based on the similarity of people’s names. This led to people being struck off the electoral register and were unable to vote.

“This is an example of what happens when automation and processes are built on data that can’t be fully trusted, or the data is used slightly out of context,” explains Lepant.


What can be done about fake data?

In the Amazon case, the e-commerce giant put in a series of measures to ensure all reviews were genuine. For instance, it provided more weight to reviewers who bought the product and created an invitation-only review programme for verified accounts.

By doing this, Amazon ensured that shoppers would trust the company again which is necessary for the business to succeed.  

But in other cases, it’s not this simple.

When AI is involved, it can be harder to rectify mistakes. The way machine learning works is that it learns from every interaction. So when Microsoft launched its first virtual assistant, Tay, in 2016, the bot eventually turned into a racist-spouting Twitter account, because it was learning from the interactions it was having with other people.

Companies need to ensure that their automated decision-making processes do not become bias or lead to the wrong outcomes.

One solution is to hire good data scientists, and increasingly, chief data officers to manage the data processes.

“We’re seeing more and more companies hiring chief data offices to make sure the data can be trusted for us and be integrated in the right way,” says Lepant.

Data is a hot topic right now and companies need to pay attention.

The Facebook/Cambridge Analytica scandal saw around 98 million Facebook users have their data used without their permission by the research company.

As a result, people are becoming increasingly aware of how their data is being used by companies.

Lepant believes that people will become more careful with their data which is why companies need to work hard to gain their trust. “Fake data could have a negative impact on the ability of organisations to capture data moving forward because people are becoming more perceptive about what they share,” he explains.

“Having consumers trust the organisation will become a key lever for businesses to build upon.”
The three ways to prevent fake data

Fake data does sound slightly terrifying but companies can ensure that it doesn’t impact their business. In fact, Accenture says there are three ways to do this.

Step one: provenance. Verify the history of data from its origin throughout its life cycle.

“If you buy data from a trusted third-party source, it is on the organisation to verify the data themselves before they use it for their own purposes,” advises Lepant.

The second step focuses on context. “In my experience, one of the key risks organisations take is using data in a different context than what it was captured for in the first place” says Lepant.

For instance, US airline company United Airlines made the mistake of making business predictions based on old data. This contributed to $1 billion a year in missed revenue, thanks to inaccurate data.

So make sure to use data in the right context and at the right time, advises Lepant.

And finally, Lepant says companies need to maintain the integrity of the data. Ensure all data the company holds is secure; providing lineage and traceability about how they acquired the data, so it can be proven at any stage.

And for you as a consumer? If you're concerned about false information: make sure to check your sources, see if the same thing is reported elsewhere, and don't believe everything you read on the internet.


Source: Evening Standard

30/04/2018