We spoke to Joydeep and talked about the several trends of the AI domain. Right from addressing the skill gaps to the problems solved by AI, he poured the best of his knowledge. Here’s the excerpt from the interview.
Data Scientists are Polymaths with knowledge in Linear Algebra, Statistics, Probability Theory, Trigonometry, Computer Science plus business domain knowledge. But they are hard to find.
One of the important ways to address the skill gaps is to adopt a Team Data Science Practice i.e., a team of Data Engineers, Machine Learning Engineers, Statisticians & Data Scientists, where each one has a different responsibility in satisfying the hierarchy of needs for Data Science. This way we can have Data Scientists focused on executing the high value, top of the pyramid use cases that deliver business value while the rest of the team members focus on a specialized task.
Besides this, it is important to have a mature Data Platform that provides a foundation to realize the Data Science hierarchy of needs like Reliable Data Collection, Data Cataloging, ETL/Feature Engineering, Experimentation with an end goal to driving AI and business value.
AI will solve the following problems –
-Hearing aids with learning algorithms that filter out ambient noise;
-Houte-finders that display maps and offer navigation services;
-Recommender systems that suggest books and music albums based on a user’s purchases and ratings;
-Medical decision support systems that help doctors diagnose
-Robotic pets, cleaning robots, surgical robots and several millions of industrial robots.
-Modern speech recognition, based on statistical techniques such as Hidden Markov Models
-Personal digital assistants, such as Apple’s Siri and Amazon’s Alexa that respond to spoken
-Machine Translation remains imperfect but is good enough for many applications.
-Face recognition at automated border crossings in Europe and Australia.
-Fraud detection in the financial services industry to approve or deny financial transactions autonomously.
-Automated stock-trading systems to automate complicated trading strategies
Deployment of AI at the edge. Rapidly decreasing costs for edge computing will see AI agents being embedded into all manner of toys, sensors, phones and myriad other devices that we use everyday.
One criticism of AI popularized by Deep Learning has been that it is capable only of pattern matching and not logical reasoning and deduction that human intelligence is capable of. Recent progress in using Deep Learning to deduce complex formulae or conversely to decipher it in terms humans understand is indicative that we will be able to reduce this gap in the future and come closer to a more generalized notion of Artificial Intelligence.
AI is also leading to great advancements in computing hardware specialized for deep-learning – and I believe we are just starting off. We spent the last 50 years evolving and refining microprocessors and it will be not a stretch of imagination to say that the next 50 years may see a similar exponential increase in computational power for AI.
With data growing in volume at a rate that will soon reach 1.7 megabytes of data every second for every person on Earth, and the existing skills gap, it’s safe to say a career as a data scientist will probably offer plenty of job/career opportunities. In order to really make an impact one should
-Start with strong foundational skills in Mathematics and Statistics that are relevant for Data Science.
-Strong skills in Programming and Data preparation are also prerequisites for practical success in Data Science.
-Access real-world data sets and attempt data mining challenges on platforms like Kaggle to hone their skills.
Cloud Computing offers everyone easy access to massive computational capabilities at a low price – and being able to exploit this has become an important skill that every aspirant must learn. Beyond this, strong business domain knowledge with ability to collaborate with several functions in the enterprise and, aligning with key stakeholders in order to connect the dots from raw data to value delivery is key to success.