Some experts have stated that the ubiquitous and potentially dangerous uses of Artificial Intelligence (AI) deem it necessary to consider sweeping regulation over the technology. It is an anthropomorphic evolutionary leap from human leadership, to automation, robotic manufacturing, deep learning and, dare I say, human technological dependence. It is also uncharted territory.
The Problem
It is precisely the technology’s pervasiveness throughout the world that presents the greatest challenge to its regulation. Machine Learning, or Deep Learning, is a result of computer algorithms. As the name would suggest, autonomous, learned, intelligence is the desired outcome of those algorithms. To say that the governing of AI is complex terrain does even not begin to take into account the diverse challenges facing potential regulatory agencies. Three of those challenges present distinct obstacles to the regulation of AI, thus making the issue of oversight a non sequitur.
The Spread of the Technology
It is precisely the technology’s pervasiveness throughout the world that presents the greatest challenge to its regulation. Machine Learning, or Deep Learning, is a result of computer algorithms. As the name would suggest, autonomous, learned, intelligence is the desired outcome of those algorithms. To say that the governing of AI is complex terrain does even not begin to take into account the diverse challenges facing potential regulatory agencies. Three of those challenges present distinct obstacles to the regulation of AI, thus making the issue of oversight a non sequitur.
What is it?
First, the broad descriptor of “AI” technology is used to describe a variety of applications which include everything from self-driving vehicles to even predictive policing and sentencing. Yet, the definition of AI, as yet, is not even universally accepted and represents diverse applications, computer code, designed to learn by itself. To think that all programmers throughout every company, across the globe would use the same techniques and language to develop this code is unrealistic. Yet, if there is no uniformity in the technology’s development, how can the process be regulated? In addition, traditional methods of regulation such as product licensing, research and development oversight, or tort liability are grossly unsuited to manage the risks associated with the technology. Therefore, new laws, and even a new legal language, would need to be written in order to define those laws that are being instituted. This exercise will take years, even if widespread agreement on that legal framework were possible.
Late To The Party
Second, it is that very time consuming exercise coupled with the rapid development and deployment of the technology that hinders its regulation. Embarking on such an effort assumes that the work of regulation is being completed before the propagation of the technology and that there is still time to contain its spread. However, such an assumption is incorrect because the technology is already in place in many echelons in the private and government sectors. The genie is already out of the bottle and any effort to suppress the technology is replete with financial and constitutional hazards. As such, these efforts must certainly be considered after the fact and therefore places the technology’s regulation beyond the scope of possibility.
Universal Consent to Govern
Third, and perhaps most important, any such effort to regulate AI must be undertaken on a transnational basis. This would require an international body to be established and recognized by every country throughout the world. That body would need to be legally empowered to act in all countries, in both the private and military sectors. If the United Nations is to be used as an example, the likelihood of this occurring is historically impossible.
To not regulate every country at the same time would place a dissenting country in an overwhelming technological advantage over the rest of the world. This is a position to which no country would willingly acquiesce.
Reality
Therefore, for all of these reasons, and more, any discussion of the regulation of Artificial Intelligence is redundant. Simply put, the regulation of Artificial Intelligence will not happen – because it cannot happen.
Simply put, the regulation of Artificial Intelligence will not happen – because it cannot happen.