Artificial Intelligence and the Future of Privacy
As AI and machine learning rely completely on data, gathered and analysed from various sources, it is not tough to imagine the risks emergence of sophisticated AI systems, poses for the online privacy of an individual.
The effectiveness of data protection has been challenged repeatedly over the last decade with inventions like cloud sharing, social media, online behavioural targeting on mobile, etc. Every time technology takes a new step towards assisting humans, it poses an all-new kind of threat to individual privacy. The latest among them is Internet of Things and Artificial Intelligence. As AI and machine learning rely completely on data, gathered and analysed from various sources, it is not tough to imagine the risks emergence of sophisticated AI systems, poses for the online privacy of an individual.
So, what are the ways in which AI can compromise the privacy of an individual? Primarily, the prominence of AI has been possible for three factors, speed, scale and automation. The analytical speed of AI, for the computation of large data sets, has been far ahead of the capabilities of human analysis. Also, the efficiency of AI for analysis of big data, and the performance of the tasks designated to it, without any human supervision have made it essential. This has also made it susceptible to ill usage by corporations and large-scale agencies, which have regular access to individual profiles.
In general, people remain unaware of the immense amount of data that can be generated by electronic devices – from smart home appliances to sophisticated computer and phone applications, with the in-built software. As a result, individuals are ignorant of how vulnerable they are to data exploitation.
The biggest problem with AI lies here. An AI Assistant can function properly only when it has access to the user’s personal data, which ranges from location and contacts and to account information and photos. Also, messages and data and all the information on a device, have to be unencrypted for the AI Assistant to function and assist the users with their daily activities. Even if a user anonymizes his personal data, once that becomes a part of the big data set, an AI can de-anonymize it, based on inferences from other data sources. Also, as more and more sophisticated AI systems are being built, identification through voice and facial recognition, across multiple devices is becoming a mandate.
This makes personal data a part of the non-personal big data pool, which can be accessed with ease, for the purpose of profiling, sorting or evaluating people, without their consent. As AI can deduce and analyse sensitive data from the non-sensitive, it can be utilized to effectively predict and infer emotional behaviour, sexual orientation, political views, and current health conditions, from seemingly insipid data like social metrics, location changes, activity logs and even, typing speed and patterns. And if the information collection and analysis done by AI do not seem dangerous enough, a recent study by Black Hat has shown how artificial hackers have completely outperformed their human counterparts, and what can be the result when AI is weaponized for stealing identities.
As the world of Artificial Intelligence continues to boom, and the world’s tendency to be completely dependent on technology keeps escalating, it is fast becoming unattainable to protect one’s privacy online. It is more of a concern for those, who are not acquainted with the necessary measures they can still partake in order to protect their private data. Though not fool-proof, one can increase their resistance to unintentional data sharing and shrink their online footprints. For starters, using VPN (Virtual Private Network) or anonymous networks, such as TOR (The Onion Router) or Freenet, while browsing the net help a user protect their online privacy. As these networks offer end-to-end encryption and hide the real IP, the data transferred during a session cannot be tapped into. Open source Operating Systems, like Linux distributions, prevent collection of data without permission, unlike operating systems by Apple or Microsoft. Lastly, browsing through open source browsers like Firefox, or browsing in Incognito mode, can reduce the online footprint of a user.
Artificial Intelligence, being a relatively recent technological innovation still has miles to go. Although, it has already been used to make substantial positive contributions to the lives of individuals, and to the world as a whole. Undoubtedly, its potential will be further utilized to tackle various evils of the society and the world. However, just like any other technological innovation, its future will be decided by the intent of the people who use it. And while none can predict its success, it might be safe to take precautions, otherwise, the tool invented to assist humans can very well lead us to our doom.