In previous articles, we’ve often discussed decision making in business, and our emphasis has been on how data and analytics help human beings to make decisions. In many areas of decision making, having a human in the loop is detrimental. Humans are expensive and slow, especially when it comes to inputting, sharing, and processing large amounts of data. In many cases, we want computers to make decisions for us: this is the science of machine learning.
There are many areas of modern life in which machine learning has an impact, and the field is being heavily invested in and studied, so we can expect that its influence will only grow over time. One market segment that machine learning is crucial for is online retail and content distribution.
Large online retailers have a problem. When a customer visits a brick-and-mortar store, retailers are able to present to them a huge array of products, just by virtue of physical presence. They can wander down the aisles, waiting to see if a product takes their fancy. While this sort of browsing is possible online, it fails to take advantage of some the benefits afforded by computation.
The ideal solution would to be able to figure out what the customer is likely to be interested in, even if they don’t know about it themselves, and present it for their consideration.
This is what Amazon does with its product recommendations, and it depends on machines having processed significant amounts of relevant data, both specific to that customer and concerning patterns of interest and buying decisions of the whole customer base. Amazon’s algorithms learn a customer’s preferences, and in doing so massively increase potential sales.
The same problem exists for content providers. How are they to surface exactly the right content to engage the interest of their users and prevent them from moving on to another provider? Some years ago, Netflix offered a million dollar prize in an open competition to the team who could help them solve the problem. When you visit Netflix, or similar services, they are able to present you with content that their systems have learned is likely to be to your taste.
There are others who are trying to make a business out of helping people navigate the information deluge and find content that appeals to them. Trapit is one such service. Based on the same technology as Apple’s Siri, Trapit is somewhat similar to a search engine, in that it takes keywords and provides relevant content, but it also uses semantic analysis and user preferences to make intelligent suggestions: a machine-led content curation system. This technology will be one component of the coming virtual personal assistant (we’re not there yet, Siri and Google Now solve parts of the problem, but not all).
As industry solves the problems that are inherent in processing large amounts of data and using it to make intelligent decisions, the areas of application for the technology will expand into many different areas of business and personal life over the coming decades.