	One of the key points continually restated by the study is that AI is widely applied in many fields and has mostly achieved its goals. The article states, “until the turn of the millennium, AI’s appeal lay largely in its promise to deliver, but in the last fifteen years, much of that promise has been redeemed.” One of the ultimate goals of AI is to be widespread in improving the quality of people’s lives, yet there are still tremendous amounts of research and development required before AI can reach this goal. While AI has certainly made significant progress in the past 15 years, I challenge that the promise remains largely unfulfilled.
	One aspect of AI that requires attention and the study significantly downplays is the amount of research still needed. As analysis of large data sets becomes more demanded, research is needed to scale machine learning algorithms on large data sets to practical complexities. One of the most important factors to applying AI in real world environments is AI’s decision making. Research is still greatly needed in developing AI that can better utilize reinforcement learning and other methodologies to make good decisions in dynamic environments. This research would have many applications, from robotics to self-driving cars. Both humans and AI inherently have limits, so another key area of research needed is how to best integrate AI with human intelligence in order to optimize AI use. If AI was to have truly delivered on its promise, the list of things needed to be researched would be much smaller.
	The study also exaggerates the scope of AI application. While AI has a presence in all fields, there still remains many areas that AI remains to be fully utilized. One of these areas is self-driving cars. The article states that self-driving cars are “leading to an entirely new urban organization.” Although they may allow for more ride-sharing and efficient transportation, cars are the world’s most widely used mode of transportation and as a result it is unlikely urban organization will change so much as to be “entirely new.” Self-driving cars are used infrequently in only select cities, and a lot of research and development is still needed until these cars become the dominant mode of transportation in the US. Another technology AI still has fairly limited application is in clinical healthcare. Though potentially life-saving and revolutionary to medicine, current AI is still a long way away from being used optimally in medicine, and a great deal of research is still required. The article also does not discuss potential safety concerns in medicine using AI, as even minuscule mistakes can be fatal to patients. Though AI has a wide variety of uses, these two areas provide key counterexamples for AI achieving its promises.
	Another flaw of the study is that while it suggests where the field of AI may be going, it does little to suggest how we should get there. The study fails to point out specific areas of research needed, as well as how to obtain the resources for such research. The study also does not examine how to overcome social inhibitions regarding AI (especially self-driving cars), perhaps a larger challenge than the technological ones. I would have liked to see the study cover these issues, as they could have provided key insights into how AI can deliver on its promises, which it is mostly yet to do.
