By Renu Jain, Dean, Institute of Engineering and Technology, JK Lakshmipat University
AI and Data Science are the known trending and transformative topics in the field of Computer Science and Engineering. The demand of AI and Data Science experts is continuously growing in healthcare, finance, marketing, education and other interdisciplinary areas. Can we say “AI and Data Science are two different branches or just two sides of a same coin”? This article explains the basics of two fields and highlights the similarities, differences and interrelations from the perspective of expertise and job profile.
The term AI (Artificial Intelligence) was coined long back in 1956. It is assumed that AI is a science that requires us to understand “Human Intelligence” and it is an engineering that requires us to develop processes and algorithms to simulate human intelligence (HI) into a machine. Though Psychologists, Computer Scientists and Mathematicians have defined intelligence from their perspective in different ways, they agree that building any type of intelligent machine requires an enormous amount of knowledge. Therefore, to build any AI system, an engineer should be able to identify the knowledge, acquire it, store it, and then design and develop efficient processes to search and retrieve the correct knowledge for inferences, decisions and further learning.
The term data science was coined in the late 20th century. Though the term Data Science sounds more like a science, I understand data science more as an engineering that requires building models and processes for digging up large data to extract the important information hidden in the data. However, to build the models and processes for a complex working system, data scientists study the science behind the system through statistical analysis, hypothesis testing, pattern and feature identification before applying any statistical and machine learning tools to get the insights of data.
Statisticians have been using different machine learning algorithms like regression, classification, clustering and other statistical tools to analyse the data for predictive analysis for a very long time in fields like weather, market, health, and business but with limited amount of data and restricted computing power. Presently, with an enormous amount of digital data and high-performance computing power (multiple core CPUs along with GPUs and vast RAM), data scientists build faster, more reliable and more accurate predictive and decisive systems using advanced machine learning models.
The availability of digital data, high computing power and performance of machine learning models on big data motivated the AI developers to build learning models where AI developers need not explicitly identify the patterns, form rules, handle ambiguities and contextual knowledge, instead system learns on its own through machine learning models trained on large datasets. The self-learning analogy of intelligent systems is rationalized with the learning of small kids who learn by watching and listening the repeated patterns. However, Data Scientists form hypotheses, get the data collected, organized and structured for analysis, create algorithms and models to answer the queries of higher management and help the organization to take the appropriate decisions. Hence, AI system developers and data scientists both model and process large data to build an intelligent system or get the information embedded in the data, filling the gaps between the two branches. In my view, tools and techniques used in data science support in the development of AI systems and these AI systems support data science in decision-making, but, in both the branches human role would continue to be important due to our critical and innovative thinking abilities and passion to attain the desired goals.
The demand for skilled AI and Data Science professionals is rich in the job market. All big companies like Microsoft, Google, Amazon, Apple, Nvidia, Uber, Cruise, etc and new companies like Numerator, Databricks, Unified, Teradata, Algorithmia, etc have big or small Data Science team depending upon the size of the company. Most of the big companies have AI jobs too like AI Product Manager, AI Ethicist, Robotics Engineer, AI consultant, etc. In most of the companies, both the teams collaborate closely to build a complete system. For example: the companies, building driverless cars have a team of AI experts who design and develop artificial intelligence systems that enable the cars to perceive, understand, and navigate their environment on their own and a team of data scientists who refine and study the data collected from sensors using machine learning models ensuring the safety and reliability of the entire system.
Let me share my experience of working on an AI project. Machine translation was considered one of the important applications of AI when the term AI was coined. Around 1994 in IIT Kanpur, we worked on a project of building a translation system to translate from English to Hindi called AnglaBharti system. It was a Rule-Based system where the work was started from scratch like creating English to Hindi dictionary, forming rules for parsing the English sentences into phrases like noun-phrase verb-phrase prep-phrase, etc. converting the parsed structure according to Hindi language and then the Hindi generator generates the Hindi version of the sentence. To develop all these modules, language experts of Hindi and English, data entry operators, AI experts and programmers having the knowledge of AI languages in addition to the senior academicians and AI researchers worked together to build a working translation system. The initial goal was to develop a general translation system from English to Hindi but eventually, a workable system could be created only for the medical domain. But after 2010, researchers could develop a good quality translation system using LLMs without going into the details of understanding the languages and the translation process.
As AI aims for learning and Data Science extracts knowledge from the existing data, it can be said that AI and Data Science are two different branches, but they seem two sides of a same coin.
