The report firstly, states categorically that the study panel had found no cause to implicate AI as a threat to humanity. The report extends its statement by passing a prediction ascertaining the safety of human future. On the flip side, a lot of business leaders and researchers have expressed their apprehensions regarding the safety of humans’ due to the unforeseeable nature of A.I. One thing that statements such as the reports’ try to prove is the predictable nature of A.I. To the contrary, many experts and including the report at other instances believe A.I to be an extremely unpredictable field/subject. This unpredictable nature of A.I not only questions the developments in this field but also sheds skepticism on the future human safety. Elon Musk has been quite vocal about the perils of extremely advanced A.I. He believes A.I. at the current pace will turn devastating self-learning machines against their creators. Deepmind, which is a start-up acquired by google, is playing a significant role in making systems, which when given a raw set of data, learn and adapt to general situations automatically. Surprisingly, these machines are not pre-programmed. The report mainly bases its assumptions and proclamations of safety on A.I. being specific to a given field. However, the projects in Deepmind are inherently based on general items hence, paving way to artificial general intelligence. Further, it is alarming to know that, google a very “information rich” corporation, has been using Deepmind algorithms to help devise various polices based on their massive user data. Hence this uncertain nature of A.I brings ambiguity to the state of human safety. Another recent example to add to this ambiguous nature is Google’s Neural machine translation (GNMT) system which has been able to establish intuition from human languages. This project, which was mostly used for language translation, developed the notion of meaning for words from a given language leading to wide spread speculation that an artificial language was being used by the system to achieve this. Understanding the meaning and developing intuition behind human language are quite different from existing natural language models which predict suitable words from data. The basic things which separate humans from machines are closely being bridged thus adding to the dense complexity of the field. Therefore, questioning the future of humans and their safety. 

Secondly, the report talks about A.I. entering the workplace gradually and the replacement of jobs being a gradual process. The report bases this assumption on historical employment data. However, historically jobs have been replaced by machines but not by very intelligent ones. The degree of intelligence of machines has grown exponentially. Machines are becoming more and more versatile. Our computers can be calculators, entertainers, compilers, and what not? Our computers can now also implement complicated machine learning algorithms with the help of libraries such as Tensorflow and Pytorch. With dropping prices of physical sensors, computational power and, memory accompanied by a plethora of data, serviceability of A.I has grown aggressively. Hence, commercial corporations will replace jobs with these intelligent, less expensive and accessible machines within no time. The moment a break-through is registered, an affordable and more accessible version appears in the markets almost instantaneously. Roomba, as mentioned in the report is a good example. Hence, factors such as lowering prices of hardware components, increasing accessibility to A.I. would not make the change in jobs a gradual but a spontaneous process. 

In conclusion, given the unpredictable nature of A.I., reports such as these cannot ascertain the human safety. Further, due to the increased accessibility and lowering prices, corporations will be quick but not gradual in replacing jobs.  
 
