Machine learning has seen significant hype in last one year. Applications range from chatbots to image searches to speech recognition to financial market prediction and more! Well the hype is all worth it. (Except may be for financial market prediction. Not may be, definitely.) ML will live upto its hype as:
It reduces friction in customer interactions. Its easier to type a query when I’m shown suggestions for example.
Sometimes the code size and complexity reduces significantly by using ML. For example to detect the gender of a person from the name one can write 1000s of rules or one can write three lines of code which learns from millions of names.
Rules are brittle. ML is smooth. Rules are brittle because they change over time and also they give deterministic results. Whereas ML gives fuzzy response and can be updated continuously.
Advances in engineering aspects of ML have been tremendous in the last 5 years. There is no significant change in underlying ML models, but their implementation has shown drastic improvements.
Software moving to the cloud and expansion API economy helped in better sharing of data required for ML