By Andrew Ledford
A long time ago, in a Galaxy far, far away and 2 B.K. (2 years before kids), my wife and I were weekly movie-goers. We would pretty much go to see just about anything. These days, 10 P.K. (10 years post-kids), the opportunity doesn’t come that often so we are pretty selective with what movie is “baby-sitter worthy.” The most recent movie we saw was Spike Jonze’s “Her” starring Joaquin Phoenix in the not-so-distant future as a lonely writer who falls in love with his Operating System (OS).
In seeing “Her” and considering my newly found field of Sociology, I began to wonder if we are truly in the dawn of a new age for our discipline. The type of AI as depicted in Her is different from the current program that “learns” your preferences on the Internet (which is slightly sinister), but rather true AI that can independently think, communicate, and most importantly reason without human influence. This is considerably more advanced than what has been discussed in previous sociology conferences such as the one hosted by the National Science Foundation in May 1993. This conference looked at AI only in terms of “intelligently searching and analyzing data.”One of my colleagues upon reading the first draft of this post suggested that I not get too “sci-fi” with my concept of social interaction with machines but I believe that is exactly the point of this concept and the necessity it entails.
Once the technology is there for AI to exist (as described earlier), the time it will take to surpass human capabilities will most likely be short considering the incredible capacity for simultaneous calculations and an extensive database of knowledge. This is not just my opinion either but a consensus of an eclectic group at the Future of Humanity Institute at Oxford University. As fantastical as the name implies, the Future of Humanity Institute has been deliberating over the existential risk facing mankind once AI “comes online.” Scientists, mathematicians and philosophers come together to work through what happens when we aren’t the smartest beings on the planet anymore. In November, I attended a lecture by Oxford University Philosophy Professor, Nick Bostrom, whom is also the founding Director of the Institute. The lecture was truly awe-inspiring when considering the term “the fate of humankind” was being tossed around like we were on the ground floor of the Manhattan Project.
As the Institute would argue, when, not if, AI surpasses human capability and becomes a “being” that can reason and interact with humans, it will be critically important to examine the interaction between the human race (now for the first time as a member of the “rational being” unit of analysis) and AI machines. To do this, an AI sociologist will have to be a stellar programmer (e.g., Python is the first language they learn and then English), and the real dilemma will then be whether it would even be possible to study machines that are vastly superior in intellect. Will they let us? One can conclude that studying a superior “being” might be similar to playing chess with a vastly superior player and wondering what their next move might be. The challenges for human sociologists in understanding the reasoning of a machine might be even more difficult. Assuming that the superior intelligence of AI will allow us to examine and study it, is it also possible that AI will begin to examine us? It is not hard to imagine that there is the potential for AI to become sociologists of the human race as well in a desire to work alongside their creators.
They will be smarter, more efficient, and most likely have an ability to eliminate bias altogether, which their human counterparts cannot do. Its quite possible that a machine would not have to disassociate itself from its own SES, country of origin or education. Would it necessarily have the characteristics of the programmer that created it? Would it be possible for the AI, assuming it does surpass human capability to then remove this characteristic altogether? There are more questions than answers with this concept as the technology still doesn’t exist but it can become quickly unnerving. It is not hard to imagine that an AI sociologist (the machine version) will be able to study human-to-human interaction better than a human sociologist will.
Back to current day— it is an exceptional time to be a sociologist with the modern tools that are already bringing down many barriers to better sociological study than was possible even 10-15 years ago. The advent of true AI likewise poses exciting new opportunities for this field. Whether we are the examinees or the examined however, AI sociology requires serious contemplation and most importantly, involvement from the sociological community in discussions on the advent of this technology, lest we find ourselves in the position of trying to play catch up with a more intelligent being. At that point, even John Connor won’t be able to save us.