Can big data save us all?

We’re in the age of big data, they said. It’ll solve all your problems, they said. But, as Katie Kenny reports, it’s much easier to exploit data for profit than social good.

Uber’s vice president of design, Michael Gough, sees his role as designing not only the company’s products, but also its reputation, and it’s clear he’s good at his job.

Around 2000 people have gathered in Auckland’s Aotea Centre for the ambitiously titled ‘‘Future of the Future’’ design symposium, featuring Airbnb, Google, Netflix, Facebook, and Uber.

Here, glasses are largely non-functional. Speakers are ‘‘visionaries’’. Their work isn’t simply design, but ‘‘storytelling’’. And Uber isn’t a ride-hailing app. It’s ‘‘an opportunity company’’.

Gough perches on a stool like a country singer, the top buttons of his shirt undone, and highlights the company’s advances into food and freight delivery, and self-driving vehicles. He shares his vision of a future with flying cars: ‘‘Imagine a world where every transportation system talks to every other system.’’

Imagine, Gough continues, a city totally connected and constantly exchanging data between, say, roads and street lights, and new systems not yet invented. ‘‘If you have all this data, you ask what else can you do?’’

If you use Uber, you’ll be aware it collects the information you provide when you create your account, as well as your location and device details when you use its services. Depending on your device settings, it’s also likely accessing your address book, tracking your location when you’re not using the app, and following your activity on other websites and services.

Not only can the data alert maintenance workers to a pothole, it can also measure the impact of major events and analyse movement patterns to help inform urban planning.

Gough says Uber can address ‘‘transport poverty’’, and its consequences, including unemployment. Lofty stuff for a software company that continues to grapple with complaints about it flouting laws and fighting with regulators.

There are many versions of this sales pitch: big data will run, and save, the world. It will combat poverty, climate change, unhappiness, and crime. But first, it seems, it will make the likes of Facebook and Google billions of dollars, and greatly improve our online shopping experience.

While technology giants have ownership and control over the largest repositories of personal data, making our lives better on an individual and societal level isn’t their first priority.

So how can society strike the right balance between exploiting the economic potential of big data, and harnessing it for social good?

In May this year, Christchurch-based Jade Software revealed it had built a machine learning algorithm, which can predict – with 92 per cent accuracy – whether a student will drop out of university, by analysing 15 years’ of student records, including grade point average, age, distance of residence from campus, enrolment history, ethnicity and how they pay for their studies.

Universities can then identify risk at an individual or group level, and offer appropriate care and support, explained Eduard Liebenberger, head of digital at Jade.

Speaking to me again, months later, Liebenberger says some public concerns in response to the tool surprised him.

‘‘People were worried about the tool being used to profile applicants for universities to, you know, get only the good ones. And that’s something we wouldn’t engage in.’’

People balk at the idea of machines making decisions about them, but that’s not the goal, he says. The machine simply makes suggestions, and trained professionals use that information to better inform their decisions. ‘‘Data is always just one side of the story.’’

As digital platforms have shifted from

making money by selling advertising to making money by mining and selling personal data, we’ve tolerated it. We’ve waived privacy for convenience every time we’ve hailed an Uber, or shared a photo on Facebook.

Clearly, these platforms have a huge advantage when it comes to harvesting data: the power of incentives. We’re less willing, to share information with non-profits, health providers, or public agencies, which in turn limits the ability of these organisations to use data for good.

By design, to protect our privacy, government data is siloed and difficult to access. But the potential of that data is huge, Liebenberger says. The same sorts of models that can predict which Netflix show we want to watch, or whether a student is likely to drop out of university, could be improving the efficiency of almost every health and social agency, he says.

Perhaps the thing that scares people most about machine learning algorithms and artificial intelligence is that these subjects are difficult to understand. Most of us feel as though we have little to contribute to the discussion.

But by staying silent, we’re allowing a small group – technologists, mainly, or executives, or public officials – to dictate how our information is used in a way that affects all of us.

David Parry, head of computer sciences at Auckland University of Technology, has spent a decade studying mistakes made by anaesthetists during surgery. He does this by using wearable trackers to detect whether an anaesthetist is moving in a way that suggests they’re likely to make an error during simulated operations.

One day soon, Parry hopes, the technology will be able to monitor and warn surgical teams in real time, and prevent needless deaths.

However, these advances have also raised moral dilemmas and he expects such close observation could be a hard sell to clinicians.

‘‘No one wants errors, but there has to be explicit understanding of what you can and can’t do with this information. Because it’s not going to be just in hospitals. It’ll be in factories, mines, any workplace where there’s risk. We’re going to have to have contracts and laws around this.

‘‘The big danger is employers will assume that, because you’ve signed up to provide this data, it can be used for anything. And that’s not acceptable.

‘‘Let’s say you have to downsize your department, well, you can’t just look at the data and drop the doctor who’s made the most mistakes.There’s potential for a lot of misuse,’’ Parry says.

Head of design at Figure.NZ, Nat Dudley, agrees: ‘‘Data isn’t benign. And the use of data to drive decision-making is especially not benign.’’

Set up in 2012, Figure.NZ is a New Zealand charity with a mission to make data more accessible. As well as publishing more than 40,000 charts and maps about New Zealand for public use, the organisation also works with schools, businesses, media, and government agencies to teach data literacy.

‘‘When you only care about profit, it’s easy to ignore the complexities of humanity, and design algorithms that use data to do just that,’’ Dudley says. ‘‘If you want to harness data for true social good, you need to be a lot more careful about bias and data quality.

‘‘It’s tempting to think that the Government should use all this data to identify how to change the country for the better, but it’s more complicated than that.

‘‘The data doesn’t tell the whole story about people’s lives, and to really understand how to improve things, you need to get out and talk to the people who are impacted. This all takes time and money, and these are complicated problems, so I’m actually relieved that Government hasn’t just jumped into using their byproduct data to make massive decisions about our future.’’

Liz MacPherson, government statistician and chief executive of Stats NZ, is custodian of the country’s most valuable database, the Integrated Data Infrastructure (IDI).

The IDI brings together data from a range of sources and removes identifying details. It stretches back to births deaths, and marriages from the 1800s, but most is from the 1990s.

Researchers can access the IDI for projects that improve the lives of New Zealanders from designated laboratories around the country.

While Ministry of Social Development and Ministry of Justice staff have done risk modelling work, more commonly it’s used for traditional research projects such as looking at cardiac risk factors and understanding the gender pay gap.

To date, the resource hasn’t been used for machine learning. ‘‘I don’t think we’ve scratched the surface of what we can use it for,’’ MacPherson says.

Does this mean we’ve been too cautious? Possibly, but sometimes nuance is vital. ‘‘There are critical decisions being made where you’d like to have a human involved, and there are other areas where you could use more machine learning and artificial intelligence to make processes just so much more efficient.’’

Although data is now ubiquitous, it’s still a slippery concept, sometimes described as a natural resource to be mined and refined, or as a byproduct like exhaust or smoke.

Sitting in a sunny Wellington office, MacPherson says data is ‘‘key infrastructure’’, like a school system or sewage network. People need to be trained in how to use it, and to follow rules.

But data is also part of a person. ‘‘Every dataset is a person, a family, a community, an iwi, a hapu¯ , a wha¯ nau. And we need to be taking data seriously from that perspective, and looking after it in a way that maintains people’s trust and confidence.’’