Machine learning techniques or translating- machine learning may have been used for years, but recently it has had an explosion in its applications. Google CEO Sundar Pichai says that “Machine learn is transformative, with it we’re rethinking how we’re doing everything.”
In the past, the success of using machine learning algorithms was very costly and there would have to be large r&d investments, but all that is changing. IBM Watson, Microsoft Azure, Amazon, and Alibaba all launched turnkey cloud-based machine-learning SaaS solutions in 2015. At the same time, startups like Idibon, MetaMind, Dato, and MonkeyLearn have built machine learning products that companies can take advantage of.
Gartner already puts machine learning at the top of its hype curve, and no: machine learning won’t replace all your employees with computers or suddenly double your revenue. But that doesn’t mean it can’t give every company a competitive advantage. There are plenty of business processes that can benefit significantly from machine learning.
So how will machine learning change the way companies operate? Read on and see some of the many ways machine learning will affect business.
Impact of machine learning on businesses
Upfront costs
Consider this: Machine learning needs training data and training data costs money. Especially training data labeled by humans.
Example: To run machine learning work for your business, the algorithm needs to see many, many examples of what it is supposed to do. If you want an algorithm to tell you if a sales leader is good, you need many examples of good opportunities and sales data, as well as bad sales leads. If you want an algorithm to mark the support tickets you need to show many examples of support tickets. If you find your algorithm for a new language, you will probably need to collect lots of examples in that language.
In some cases, a company may have training sets at home. For example, a lot of qualified or disqualified leads. But let’s say you haven’t marked each of your support tickets as they come in throughout the year. You would need to have people – whether at home or in bulk through a data enrichment platform – label the tickets. The machine then looks at the judgments and starts finding connections and patterns that it can learn.
Much lower ongoing costs
Machine learning is much cheaper and more efficient than people when it works well. The downside is that it often works well in 80 percent of cases and poorly in 20 percent of cases, and reducing the error rate to 20 percent is difficult, if not impossible.
But even an 80 percent precise algorithm can save you a lot of money because good machine learning algorithms know where they are needed and where they are most likely to have errors. Smart companies take cases where the algorithm has high confidence and uses those directly when issuing low-confidence cases for humans.
The banks have been doing this for years. When you place a check on an ATM, an algorithm tries to decipher the numbers in the check. If you have really sloppy writing or the ink is stained the algorithm passes the task to a human. This design standard saves a lot of money banks while preserving a very high level of accuracy.
- Your costs will fall over time
A huge benefit of machine learning is that it can become part of its variable cost at more than a fixed cost. If you use humans to handle cases where the algorithm is struggling, you are creating the perfect training data to feed into your algorithm. This is a well-studied technique called active learning – it turns out that the data labels collected training in cases where the algorithm has a low confidence level helps the algorithm learn much, much more efficiently.
As the algorithm becomes increasingly accurate, the unitary economy of your business process becomes better – and as machine learning becomes able to handle more cases, expensive humans are called only in more difficult situations, the rarest. This means you use the best of both human intelligence and the machine together: leveraging the speed and reliability of computers for easy judgments and the fluency and knowledge of humans for the most difficult. And if that kind of thing, it’s because it is.
