Machine Learning: the most cutting-edge technology in today’s world. We hear about it in the news, in our businesses, in social media and through seemingly every big corporation. For many, this oversaturation means that the field of ML is on the decline, and that in 5 years we will have exhausted its potential.
What if I told you that Machine Learning has been around for 60 years?
Machine Learning as a term was coined by Arthur Samuel back in 1959, the concept of a learning machine was proposed by Alan Turing in 1950, and (here’s the kicker) the first neural network was built in 1951. So why all the commotion now? If ML and AI have been around for so long, is it finally time to move on?
The first thing that has to be established is how broad AI and ML as terms are. There are so many algorithms, models, fields and areas or research that can all be categorized under these fields, so it’s important to understand advancements in a certain area don’t equate to advancements in others. AI is an area of research, just like psychology; we don’t evaluate the future of psychology as a field every time someone discovers something or when businesses start applying new psychological discoveries, so why should we do that with AI?
Computer scientists and mathematicians have been designing algorithms since their inception as a career. The trendline function on Excel? That’s an ML algorithm discovered by Gauss in 1809. The real reason AI is all around us today is because the world is finally ready to feed the hungry baby what it wants: data and processing power.
The first application of Reinforcement learning (the algorithm that led to the well-publicized victory of Google’s AlphaGo in 2016) was not done in Python, Java, or even on a computer; it was done on 304 matchboxes! We finally live in a world today where ML algorithms can train themselves on billions of samples, and therein lies the recent upshot of AI to the forefront of the world. These algorithms are finally breaking free of its academic shackles and solving real-world problems in the corporate world.
AI is nowhere near the end of its life though. AI is, at its core, a tool that is only now beginning to be leveraged to solve problems. In the academic and corporate world, new research in CNNs are helping self-driving cars see; LSTMs are transcribing and composing music; and it took me a grand total of 15 minutes to leverage Scikit-Learn, a Python library, to detect whether or not a patient had breast cancer to 97% accuracy.
There’s so much new research being done by the smartest mathematicians, computer scientists and data scientists across the world. Just 12 hours ago, MIT Tech Review posted an article about a robot teaching itself to walk. AI in its diverse, varied, nuanced, cross-pollinated, and its ever-changing goodness isn’t going anywhere. There’s still so much left to do!
Contact me at dron.h.to@gmail.com if you’d like to talk!
