OPINION: There are too few protections for individual privacy in the artificial intelligence marketplace
We find ourselves in uncertain times regarding the continual development of machine learning technology. Artificial intelligence and its applications have come a long way in an extraordinarily short time. As it stands, there are too few protections for individual privacy in the emerging AI marketplace, and tech companies are left to regulate themselves in the interest of commercial privacy.
To right the ship, the U.S. federal government and state governments like Idaho’s should focus on protecting citizen privacy and scale back the accelerationist approach to AI that they have historically favored.
The U.S. regulatory environment for artificial intelligence has been lackadaisical. Falling in line with most regulatory efforts in technology, artificial intelligence has been left relatively untouched, reflecting a prioritization of innovation over user privacy. Accordingly, recent developments in the AI space have come with privacy concerns.
For example, LinkedIn has recently gone under fire for accessing its users’ data, without express permission, to train AI models. There is a button in the LinkedIn settings which allows users to opt out, but the steps required to do so are not immediately obvious. It is not in LinkedIn’s best interest to train its models by asking users to opt in, and without any regulations regarding how they should train their models, they do not have to worry about accountability to anyone but themselves.
In the absence of regulation, big tech firms are left as the sole arbiters of what is right and wrong in the emerging AI market. Individual privacy is sacrificed for the sake of continual and accelerated development of AI.
On the other side of the fence, tech firms are using corporate privacy as a shield against potential regulation. This will surely be furthered by members of the recently established National AI Advisory Committee. Some of those members have maintained their primary employment as executives of major tech firms in the AI space. Those members include Miriam Vogel, president and CEO of Equal AI; James Manyika, senior vice president at Google; and Swami Sivasubramanian, vice president for Data and Machine Learning Services at Amazon Web Services, just to name a few.
This is not to say that patents and intellectual property do not serve an important purpose for furthering innovation, but we must be willing to sacrifice some corporate protection for the sake of individual privacy and safety. Governments should ensure a modicum of transparency in AI training and development so that we can preempt any major safety or privacy issues that come from new technologies.
The federal government and various state governments like Idaho’s would be wise to take note of the European regulatory model.
The General Data Protection Regulation is a European Union law that protects the privacy of individual Europeans. This regulation covers all data related activities that take place within EU countries.
Three aspects of this regulation are particularly important to highlight. First, organizations are required to have explicit consent from individuals to process their data, this solves the LinkedIn problem. Second, organizations are required to maintain transparency in how they use the data that they collect. Finally, organizations processing personal data must ensure compliance through continual oversight.
This is a comprehensive method that the EU employs to protect its citizens from predatory data gathering practices. Regulations like this will become increasingly important as we inevitably become more dependent on AI moving forward.
• • •
Spencer Reed is an undergraduate political science student at Boise State University with an interest in policy research. Originally from Thousand Oaks, California, Spencer is passionate about understanding the complexities of public policy and its impact on society. His academic pursuits aim to contribute to ongoing discussions around effective governance and social change.