AI and the Future of Privacy

Artificial intelligence is one of the most disruptive technologies, and though it’s only in its infancy, we can be sure that it will revolutionize digital transformation and the entire world.

But as they say, with great power comes great responsibility, and given that AI is capable of collecting and processing huge volumes of customer data, it’s only logical to ask ourselves whether our privacy is protected.

Brands heavily rely on AI and its subsets machine learning and deep learning to help them target their audiences with the right marketing messages as well as predict customer behavior. In other words, machine learning identifies different hidden patterns in our online behavior, combines these findings with what other consumers from our demographic have revealed and makes fairly accurate assumptions about our interests, preferences, and needs.

The thing is that we don’t have to directly reveal any sensitive, personal information – our digital footprint is what gives us away: every website we visit, every social media post we like, or a search we conduct on Google is being taken into consideration when it comes to building our customer profile.

So, are AI and other latest technologies such as voice and facial recognition eroding our privacy and turning us into a moving target?

The Concept of Privacy 

In our information-based digital world, the notion of privacy has become much more complex than it used to be, say, 50 years ago when it mainly concerned physical, bodily privacy.

It’s much more abstract these days, and it’s hard for an average person to imagine that by clicking the “accept” button in numerous end-user agreements they actually allow different companies to collect, store, analyze, and use their information. Who could blame them when all these agreements consist of pages and pages of fine print written using somewhat incomprehensible legal lingo?

And even if the idea that their personal information ends up in a huge database crosses their mind, they aren’t aware of potential implications and threats they will be susceptible to.

However, a couple of data-breach incidents that happened recently shed a different light on this issue.

The Facebook-Cambridge Analytica scandal which compromised the information of 87 million people was one of the biggest wake-up calls for the general public. Average social media users finally began to understand that their anonymized data can easily be mined from a huge database, de-anonymized, and used for different purposes without their consent.

And these purposes go well beyond marketing, advertising, improved customer experience, and product recommendations.

For example, 70% of companies use social media to research their candidates during the hiring process, while 48% of employers use social media to check on their current employees, while 34% of employers even punished or fired their employers over something they posted online.

However, China has taken the Big Brother concept to a whole new level by setting up the so-called “social score” which will rank citizens based on their behavior and punish or reward them depending on it. This program, which raised flags with Human Rights Watch, is expected to be fully deployed by 2020, while millions of people have already fallen victims to the blacklist and denied the right to travel.

So, clearly, despite making our lives much easier, AI poses a threat in the sense that it can be used to monitor and track people, while facial recognition can put all of us under surveillance.

Is It All That Bad?

All these downsides of AI and other sophisticated technologies can be just growing pains as humanity is trying to figure out how to best implement them without sacrificing their immense potential.

This is a huge challenge if we bear in mind that our world has become increasingly interconnected.

IoT, which has taken the world by storm, spins a digital web between devices, machines, people, animals, items, you name it, and helps us, for example, control our computer, car, or home remotely. You don’t have to be a rocket scientist to conclude that such a massive network of interconnected devices creates a gargantuan amount of data that can be compromised.

But, despite concerns and numerous worst-case scenarios, this particular technology has the power to reduce carbon emissions, prevent water and food waste, as well as improve our health, safety, and accessibility.

Similarly, chatbots collect customer data and analyze it in order to create a knowledge database and improve customer support as well as personalize the customer experience by providing the best possible answers. It’s the data you willingly share with brands that makes it possible for them to tailor the best offer for you and provide you with superb service. It’s what allows them to understand you, your needs, pain points, and concerns, and find a way to meet your demands.

Similarly, AI in the healthcare industry raises numerous privacy concerns over the use of customer data, and yet the very fact that millions of people agree to share their personal information helps healthcare providers and companies improve their products, offer more accurate diagnostics, and accelerate medical research. For example, by giving your consent for your heart rate monitoring device to share the information, you can contribute to a study researching heart disease and ways of preventing it.

But, again, insurance companies are trying hard to get their hands on all this data in order to assess your health risks – your lifestyle is sedentary, your heart rate goes wild while you’re driving to work, and you sleep only a couple hours a night, and just like that the cost of your insurance goes up.

Is it possible to keep these beneficial sides of AI and at least minimize the privacy issues?

The GDPR is one way of achieving this, but this regulation also imposes some limitations which will significantly impact companies’ ability to innovate with data as they won’t be able to collect new information before they understand its value – and for many companies, it’s sometimes impossible to reach that conclusion in advance.

Potential Solutions

In a nutshell, researchers are trying to combine cryptography and machine learning in order to make it possible to learn from data without actually seeing it.

This anonymity will shield the end-users and allow companies to take advantage of their data in a more ethical way.

Here are some technologies that are currently being developed:

• Federated learning. It’s a decentralized AI framework distributed across millions of devices, and it allows scientists to create, train, improve, and assess a model based on a certain number of local models. As companies have no access to user’s raw data as well as an option to label it, this technology, which is a hybrid of AI, blockchain, and IoT, safeguards users’ privacy and yet offers the benefits of aggregated model improvement.
• Differential privacy. Different applications, such as maps, collect individual users’ data in order to make traffic predictions and different recommendations. As it’s currently theoretically possible to identify individual contributions to these aggregated data sets and expose the identity of every individual contributor. Differential privacy brings a hint of randomness to the entire process, thus making it aleatoric and preventing the possibility of tracing back the information and discovering who individual contributors are.
• Homomorphic encryption. This advanced technology allows machine learning algorithms to work on encrypted data, which means that it prevents access to sensitive information. The data can be encrypted, sent to analysis on a remote system, and the results will be returned in an encrypted form too and unlocked by the unique key. All this can be done without compromising the privacy of users whose data are being used in the analysis.

As AI, machine learning, IoT, and other advanced technologies that leverage data (some of which is highly sensitive and classified) are the future of our growth, the issue of privacy has to be dealt with very carefully as the stakes are high for both parties – users and companies who need their data.

Although it’s essential to order this field and protect people from having their personal information exposed and their privacy jeopardized, imposing strict and rigid limitations will result in hindering the development of AI.

Guest Post by Michael Deane who is one of the editors of Qeedle, a small business magazine. When not blogging (or working), he can usually be spotted on the track, doing his laps, or with his nose deep in the latest John Grisham.