The increasing availability of large data sets and access to immense compute power has seen interest in machine learning blossom. Machine learning is being used to tackle complex business issues and automate decision making. Machine learning is in contrast to previous approaches to AI in which experts developed algorithms.
What is it?
Machine learning is a subset of artificial intelligence, where computer algorithms will identify patterns in the data and apply those patterns to make predictions.
It is widely used today, and is at the heart of things like Netflix’s recommendation engine, Tesla’s self-driving car and speech to text applications.
The explosion of interest in machine learning resulted from the ready availability of large data sets which can be used to train models and the emergence of cloud computing, which has made access to vast amounts of computing power on-demand.
One area of machine learning that has proven popular recently is deep learning — an approach that uses significant compute power to build so-called neural networks, which act almost like a brain.
What’s in for you?
Successful deployment of machine learning can enable your enterprise to make predictions, and automate some decision making. For instance, you might introduce a chatbot to handle some basic customer interactions, such as answering common queries.
You might also use machine learning as part of innovative new products or services. There’s been intense interest in ML in the automotive industry around the development of self-driving vehicles.
Used wisely, ML can reduce operational costs, improve your bottom line, increase your workforce productivity and enable you to make better decisions faster.
What are the trade offs?
Machine learning is highly dependent on the quality of training data. If there are flaws in your data set, the models may learn to make poor decisions or can drive unethical outcomes by exploiting the inherent bias in your data sets.
Even more concerning, many current data sets have been shown to contain bias. If these are used to train models, you may end up making discriminatory decisions. To counteract this risk, there’s increasing interest in being able to understand and explain how machine learning systems arrive at their decisions, as well as having diversity in the teams working on these initiatives.
And while it's becoming easier to access, train and deploy machine learning models, hiring data scientists — who you’ll need to ensure success — remains expensive.