The Dos and Don'ts of Machine Learning
by
Marla Rosner
Machine learning is coming to disrupt industries whether you like it or not. The important thing to do now? Make sure you’re ready—and that means understanding what sort of role you can expect AI to play in your company.
The don’ts
To know where to implement AI, first you have to understand what machine learning doesn’t do. AI is not simply the next automation tool, and if you treat it as such, you’re wasting its potential and your own time and money. You don’t need AI to solve linear, transparent problems that operate by a clear set of rules; an Excel spreadsheet can do that. Don’t invest in machine learning for processes where the equipment you already have can do the job.
Of course, at the other end of the spectrum are the problems AI can’t yet solve. For all the fretting about the machines taking our jobs, there’s still plenty that no current program can do. Artificial intelligence can’t replace human creativity, intuition, and critical thinking.
For AI to be successful, the rules of the problem don’t need to be well-defined, but the problem itself must be. AI can’t tell you what makes your customers happy; only present you with the data to figure that out yourself.
The dos
So what do you use machine learning for? A great deal, if you want to weather the coming changes.
The sweet spot for AI lies in problems and procedures that are done frequently and consistently, but can’t be accomplished simply with rote logic. The key is to focus on processes for which your company already has a great deal of data, because that’s what the AI is going to learn from.
A good example is approving or denying an application for a loan. That’s a procedure that requires a deeper level of understanding than the average automated task. However, it’s also strongly rooted in the ability to sift through large amounts of data—and that’s something computers are better at than humans. What machine learning can do is find patterns and extract value in vast amounts of data, with a sophistication that no other technology can match.
By industry
This capability can be applied to industries across the board. Service and retail businesses can use AI to understand and predict consumer demand and personalize customer experience. Manufacturing can use machine learning to monitor equipment and detect potential issues before they arise. Even agriculture can make use of AI to better decide what crops to grow, where, and when, allowing for more sustainable and efficient growth.
These are the roles machine learning is now primed to fill in just about any company. Make sure you clear a space in your business, because only the companies that are ready for the coming of the machines will benefit.