In the “One Hundred Year Study on Artificial Intelligence (AI100)”, Stanford University, the panel describes a multi-layer transformation of human society through progress in Artificial Intelligence. They deny the possibility of runaway AI taking over the world and maintain that the benefits of AI exceed the negative aspects of it [1]. While the amount of wealth created and progress made by AI to date is undeniable in fields such as search and recommendation systems as well as deep learning, the majority of this report focuses on potential future progress. The article speaks as if in just 13 years we will see major industries such as healthcare and education being radically transformed. In reality, it seems that the only field close to being radically transformed is self-driving cars, in which we already have started to see self-driving cars on the roads [2]. No doubt self-driving cars in themselves will transform several aspects of society namely how people spend their times and how far they live from work, but that is not enough to make life significantly better and bring economic prosperity. The overall tone of the article is that of one in which the future is automatic, and major fields will transform themselves over time. This automatic property of the future of AI must be challenged. Human agency seems to be underrated significantly in the radical improvement of these domains [3]. From a technological perspective, a lot of the systems described in [1] are doable, however, as the paper itself describes, a number of challenges are encountered in each of these fields. In healthcare, a regulatory overload from the FDA and privacy protections lead to progress being severely stalled [1]. In education, archaic institutions such as Universities, which have been unable to improve their age-old lecture-style classrooms, cannot be expected to accept and push AI technologies in classrooms. While we have seen the development of MOOCs, it has not been enough to supplant Universities and offer a new wave of education to growing populations. Something qualitatively different might be needed for Education and AI cannot be the go-to solution to education. In the domain of robotics, we find service bots killing themselves on accident due to imperfect systems [4]. The article [1] suggests that cities will have robots for public safety, but security bots can’t keep themselves alive in 2017, how can we expect them to keep us alive in 2030? We must be careful to not overstate the possibilities of AI lest we might lead ourselves into yet another AI winter, which is not new to the field. We have had a number of AI winters before, and this is a story we’ve heard before: AI is going to change the world, it is only a matter of time [5]. In fact, it seems that the reality is more close to: AI is going to change the world, but only if we make it happen. The current state of AI is overstated, however the future potential is not. It is simply framed in the wrong manner. We must not underestimate the role of human agency in building the future and we must take responsibility of building safe, beneficial and powerful AI systems. AI has the potential to lead humanity into a utopian surplus society, in which people are freed up to do things that they want to do instead of things they have to do. However, incremental improvements and overly optimistic predictions won’t lead us there. We must think of radically new ways to build intelligence to bring the future closer.

References:
[1] “One Hundred Year Study on Artificial Intelligence (AI100),” Stanford University, accessed August 1, 2016, https://ai100.stanford.edu
[2] “Steel City’s New Wheels”, Uber, accessed September 10, 2016, https://www.uber.com/blog/pittsburgh/new-wheels/
[3] Peter Thiel on Stagnation, Innovation, and What Not to Call your Company
Mercatus Center - https://medium.com/conversations-with-tyler/peter-thiel-on-the-future-of-innovation-77628a43c0dd
[4] Did this robot commit suicide?
Yaron Steinbuch - http://nypost.com/2017/07/18/did-this-robot-commit-suicide/
[5] AI winter
https://en.wikipedia.org/wiki/AI_winter