Here’s How Machine Learning Is Disrupting Business Processing
It’s hard to believe, but the machines are among us already. “Machine learning” — a phrase you’ve likely heard a lot about put possibly dismissed — is here, today, and already doing faster and more accurate work than you or I are capable of. But what this actually looks like in practice might be a little bit mysterious.
So let’s take a look at some of the specific and, heretofore, human-dominated business processes that artificially intelligent machines are disrupting in a huge way.
Brand Exposure and Marketing
Applying artificially intelligent algorithms to brand exposure and marketing was always an obvious fit in hindsight. All of the major tech companies do it now — and it’s either because of their “anticipatory” business model that they’ve come to hold such dominion over our lives. Apple and Google deliver personalized results based on the internet videos, television shows and movies we watch, the news articles we read, the music we listen to and the places we visit regularly. Amazon does all that, and will soon do it with groceries, too.
The point is, machine learning is how companies with millions of customers and billions of dollars in annual revenue can keep track of it all and still make what they do feel immediate and relevant to each customer under their wing. And if works for them, it can work for small businesses, too, as we’ll see more about in a moment.
Customer Service and Retention
Customer service is expensive and previous attempts at automating the process, including pre-recorded phone menus, were met largely with frustration and confusion. Machine learning can change all that — plus make the process eminently more affordable even for small and medium-sized businesses.
Applying machine learning to customer service can leverage both historical customer data — including purchase dates, warranty coverage, preferences and other points of interest — and natural language heuristics to provide customers with accurate and tailor-made results, whether they’re asking general questions or inquiring about their specific accounts and orders.
And it’s working, too. Already, 44 percent of customers in the U.S. have indicated a preference for intelligent “chat bots” instead of human beings. Impersonal? Maybe — but it’s certainly convenient for all parties involved.
Like customer service, hiring is an expensive and time-consuming process. When polled for research, a strong majority of recruiters indicated that winnowing candidates down to a shortlist is the hardest part of their jobs. It’s clear they care about making the right choice — but how can you show fairness to every applicant at a time when corporate job openings can draw hundreds of applications from all across the country?
Machine learning provides a promising, albeit still imperfect, solution. To begin with, algorithms can remove applications which do not reference required minimum credentials and skills. It can also automatically remove applications with typos or which use (or do not use) chosen keywords.
It’s also true that the world of gainful employment these days is a little less conventional than it used to be and a little less fixated on traditional labels and “hard skills.” Having machines sift through resumes for keywords might do more harm than good if that’s as far as things go.
So we need to apply machine learning in a way that doesn’t reproduce human biases. Unilever has already put this into practice and claimed a victory in the fact that 80 percent of their machine-screened applicants are offered jobs. In other words, using algorithms followed by the “human touch” of selecting finalists for in-person interviews yields a better quality of applicant, on average, than screening them only by hand.
Supply Chain Management
Of course, intelligent machines aren’t just able to look after their own innards — they also show us ways to do our own jobs more efficiently than we ever could.
In the supply chain, it’s hugely important for a large number of parties and steps in the process to communicate well with one another. Think about the complexity of coordinating something like a refrigerated vaccine delivery or transporting components of larger machines across an ocean for finishing. Even moving packages from one side of a warehouse to another can benefit from automation and other machine learning processes.
By applying sensors and a variety of other data-gathering tools to product handling equipment, warehouse systems and freight docks, we can remove human error from processes like maintaining optimal environmental temperatures and positioning warehouse conveyors automatically as shipments come and go.
There are lots of jobs tied up right now in maintaining processes, machines and infrastructure of all sizes and kinds. Think about the time we spend monitoring and tuning-up production equipment by hand to make sure it keeps working without downtime. Think about the work it takes to retrieve a roller coaster car from the top of a lift chain or a broken-down train engine in the middle of nowhere. So much of our maintenance so far has been reactionary — and, in the case of trains and roller coasters, potentially deadly. At least one recent high-profile Amtrak crash was allegedly due to a miscommunication between a functional system and a system that was offline for maintenance.
You generally don’t take your automobile to the shop for anything less than a burning smell or an obvious malfunction. But soon, nearly all of the machines in our lives — starting with the business-centric ones — will be able to look after their own maintenance and signal the appropriate parties before something tragic or inconvenient happens.
However you intend to apply it, and no matter the stakes you deal with, it’s clear that machine learning has already made itself useful in operations both large and small. Maybe your company will be next.
Megan Ray Nichols
STEM Writer & Blogger
Did scientists discover another planet? Invent a new gadget? Reverse climate change? Subscribe today get the latest news.