Data Science v/s Artificial Intelligence

Original article was published by edwig noronha on Artificial Intelligence on Medium


Data Science v/s Artificial Intelligence

……. a causality conundrum!!!

I had a fight. I argued in very loud tones with a friend of decades. Our discussion started out casually, discussing the fields of Data Engineering, Data Science and Artificial Intelligence, as you often do when you are in IT. What triggered the devolvement of a pleasant conversation to a point of breakdown was our inability to see eye to eye on the relationship between Data Science — DS and Artificial Intelligence — AI. This difference of opinion was further catalysed by the presence of alcoholic beverages and My own hubris. My friend that works as a Data Scientist had confidently declared that DS was a broader field and that AI was a niche subset therein. Counter argumentative that I am, I just could not let that statement fly and authoritative that I am by virtue of my own assumed intellect, I had to correct her. I took the opposing viewpoint that AI is a broader all-encompassing field that encapsulates DS.

To justify her stand my friend advanced the argument that levels of intelligence are nothing but higher upon higher order derivatives of data. Data is intrinsic to the process of building intellect. According to it’s Wikipedia definition, “Data science is an inter-disciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and insights from many structural and unstructured data”. The compounding of knowledge extracted from data leads to a more intelligent entity. The advancements in the data-centric processes of DS and their subsequent practical application for knowledge extraction consequentially manifests as AI in machines. As DS works to solidify the methods and processes for handling the foundational basis of AI, i.e. data, it is a broader field.

Intelligence driven by data??

To me it seemed intuitive that human intelligence would have a broader scope that could be focused to have implications on the handling of data. As AI seeks to mimic the cognitive abilities of a human mind the same would hold true for AI and DS. In my initial exploration to counter her stand I tried to challenge the assumption that all the different types of intelligence as posited by Howard Gardner are intrinsically data-centric. If there was even a single form of intelligence that was not driven by data, it would help me establish that intelligence extended beyond the purview of data. It appeared that it can be successfully argued that every type of intelligence posited manifested due to the underlying stimulus of data. So far, I was losing my case and her statement seemed to be holding true.

Howard Gardner’s proposed multiple intelligences

Not one to back off so easily, I started on a different rational tangent to advance a counter. If we were to reverse the derivational process of intelligence and reach the lowest order, then at that level would the least inferential ability of an intelligent agent stand exclusive of the least datum required to catalyse the process. In an untrained agent, does the ability to process the first instance of input data exist irrespective of whether that input is realised? Now it became a deeply philosophical conundrum that was not very dissimilar to the age-old causality question: “which came first: the chicken or the egg?”. We sought to establish if data was the cause that effected intelligence or if intelligence was the cause that effected data. Also, is data even data if it is not identified as such by an intelligent agent?

While it is true that intelligence only manifests and becomes tangible when it is validated by a very physical concept such as data, is it necessarily a physical concept itself? If it was, then there would have been a precise scale to measure intelligence which there is not. There exists things or phenomena. The empirical, representational and other such constructs that make these things physically tangible are data. The methods of creating and refining these constructs would be a scientific process. The ability to undertake a scientific process, either simplistic or complex, would be intelligence. This was my very rudimentary way of organizing the concepts we were discussing in our conversation so far. This approach of using general intelligence helped me establish that data and intelligence are mutually dependent for tangibility and introduce doubt in the assumption that intelligence only exists in its physical manifestation on account of data. The possibility of the existence of intelligence beyond tangibility elevated it to an almost meta-physical realm.

Intelligence driven by an almost meta-physical spark??

Did you see what I did there? I used a form of Jesus-smuggling argument to blindside my friend because you can never win with God! And sure enough, she relented. I also found out that for the sake of upholding my super elevated ego it was not beyond me to use a form of the Jesus-smuggling argument to win a debate. My argument was won because I extricated Intelligence, using it analogously to represent AI, from the constraint of the physical. However, in taking it to a place where it is not physically provable, I may have left my argument too open-ended. Formally speaking, regarding practical applicability my friend may have been right about the matter. Colloquially speaking, I was not wrong from what I understand is the current ongoing discussion between the more involved minds in these regards. DS is a relatively more formalised field, while AI is still rather arbitrary. And that may very well be the entire reason for this juxtaposition.