On page 43, the article5 lists three policy recommendations. The first proposes that all government officials, both local and federal, be trained to a level of expertise in AI. I disagree that this is the best route to optimize the societal benefit to harm ratio of AIs development for multiple reasons. Firstly, the proposal implicitly assumes that politicians would principally use their new capacity for purely benevolent and non self-serving purposes. This runs contrary to how most politicians act. In a world where a large majority is ignorant to most AI-related matters, this portends undesirable future outcomes. It is nave to think a politician who understands how to leverage AI would only use that knowledge for the common good and not abuse such a technological advantage. It is natural to envisage that the first cohort of politicians armed with AI mastery would be able to predict outcomes and manipulate them in a way that the vast majority of voters would not understand. From the perspective of the public eye, this is a black box to be avoided as it has potential to perpetuate inequality and subvert democracy.  

Additionally, a politician seeking AI expertise is an unnecessary endeavor that would require a drastic change in how they spend their already jam-packed days. Due to shifting a large portion of politicians finite working hours to strictly AI, politicians would likely lose competency and/or simply not have the time to address issues in other, potentially more pertinent, realms. As an organized researcher who devotes their working hours to neural network modeling does not have ample time to discuss geopolitics enough to run for office, the politician does not have time to cultivate expertise in AI. Sometimes politicians already struggle fulfilling current job descriptions.1 2 3 4 This is not to discount the importance that politicians, or any citizen for that matter, understand the ramifications of a technology as powerful as AI. But one need not understand the complete inner workings of a machine or an algorithm at expert level to make a valid prediction about the value or harm it could bring to society. 
 
Instead, my proposal is to encourage all citizens, including politicians, to increase their level of awareness to future technologies (with AI at the forefront) and the existential problems they pose. The principle objective would not be to teach politicians about Bayesian Reasoning and deep learning (although it may be a beneficial derivative effect) while leaving the rest of the public in an even darker relative position. Rather, I would amend the policy to promote public conversations regarding:

1. what the underlying problems and potential remedies are when an algorithm builds its heuristics rooted in discriminatory demographics. 
2. what jobs AI really can (and will) replace, and what a future best-case scenario for those currently employed in those jobs could be. 
3. the meaning people derive from their employment. 
4. scientific literacy and its importance. 
5. awareness of existential risks. 

The masses need not understand how AI does what it does. The more pressing matter calls for an honest public conversation about the probable ramifications of widespread AI growth. In order to maximize the benefit to harm ratio that this powerful technology brings, the AI community should make as strong an effort as possible to include the public at all relevant stages as to current AIs capacities. In conjunction with an appraisal of societal goals and democratized decision-making power, this will create a feedback loop that opens up the optimal path to leverage AI for the benefit of the majority.  

Word Count: 590 

1. Condon, S. (2009, June 24). Will Congress Read Bills Before Voting? Retrieved September 12, 2017, from https://www.cbsnews.com/news/will-congress-read-bills-before-voting/ 
2. Innes, W. (2014, December 15). Members of Congress Dont Believe Its Their Job to Read Bills They Pass. Retrieved September 12, 2017, from https://ivn.us/2014/12/15/members-congress-don't-believe-job-read-bills-pass
3. Kolb, C. (2017, July 03). Sen. Rand Pauls Read the Bills Act  Is it really too much to ask members of Congress to read legislation before they vote on it? Retrieved September 12, 2017, from http://www.foxnews.com/opinion/2017/07/03/sen-rand-pauls-read-bills-act-is-it-really-too-much-to-ask-members-congress-to-read-legislation-before-vote-on-it.html
4. Schlesinger, R. (2017, May 05). Not So Dirty Little Secret. Retrieved September 12, 2017, from https://www.usnews.com/opinion/thomas-jefferson-street/articles/2017-05-05/trumpcares-not-unusual-members-of-congress-often-dont-read-bills 
5. Stone, Brooks, Brynolfsson, Calo, Etzioni,, Teller. (2016). Artificial Intelligence and Life in 2030. Retrieved September 9th, 2017, from https://ai100.stanford.edu/sites/default/files/ai100report10032016fnl_singles.pdf
