AI in Business: Getting Started
In my previous blog, I discussed the practical reality of machine learning and other AI technologies that are no longer bleeding edge. It's time for businesses to stop thinking of AI as science fiction or a technology only an organization with Google-scale resources can tackle. Practical AI for every business is possible today.
Consider the field of business intelligence: technologies and processes for extracting meaningful information from your business records. Important, but not exotic. Every business is, or ought to be, looking for ways to get more value out of its operational data, its customer data, and any other data it can legitimately get its hands on.
One change I expect to see is that BI will be driven less by human hypothesis setting and more by machine learning. So far, it’s not quite true that you can point a machine learning algorithm at a data set and expect it to make sense of that data on its own. Most machine learning software is still directed or “trained” to look for general patterns, although it can explore those patterns more quickly and thoroughly than would be possible if it waited for human direction at every stage of the analysis. Unsupervised machine learning software that requires little or no direction is still a technological frontier, but I think we’ll see it sooner than many expect.
Applying machine learning and other AI technologies to business intelligence and operational analytics is a strategy I believe any sizeable company should be pursuing today.
Start with some data set for which BI techniques produced significant insights in the past, but nothing remarkable lately. Maybe the problem is that BI tools are best at telling us what we already know. And before we even get that confirmation, we have to go through the process of normalizing our data, which is a fancy way of saying we throw out what our preconceived notions tell us is unimportant. With an AI approach, we can apply machine learning to all of the data in search of insights we never would have anticipated.
Google, IBM, Amazon, Microsoft and others have introduced AI as a service products you can access via APIs or drag-and-drop app builders. These AI as a service products have already been trained to perform common tasks, meaning you just add your own data. Feed the Google Cloud Vision API images or a video stream, and you can have it send back text extracted from those images, recognize logos and landmarks, or classify objects within the images. Or feed your sales data into IBM Watson and see if it identifies patterns or produces predictions that your BI team missed, using more conventional tools.
Learning to use these services effectively will still take skill, but not the same level of skill required to develop novel machine learning applications. Where the need to hire or train personnel as AI specialists becomes more acute is if you see a unique opportunity to embed AI in your own products or services. You should be thinking about that, too.
One of the things we’ve tried at RingCentral is applying machine learning to analyzing network data patterns and predicting congestion and potential network failures so we can route around them. It’s allowing us to prevent problems before they can occur. Groundbreaking? Maybe not, but certainly a practical application.
What will be the equivalent for your industry? I’m not sure, but I’m sure it will be something. For most of us, the biggest danger of AI is not that it will take our jobs, but that it will make our jobs irrelevant by transforming our industries in ways we failed to anticipate. Failure of imagination is the biggest hazard.
My main goal with this article was merely to convince you to start including AI in your plans. You’re not wrong to be skeptical of overblown promises, but do not let that stop you from experimenting and exploring the real value for your business.
That doesn’t necessarily mean you need to hire a team of Ph.Ds to do AI research, but it might if you see an opportunity to embed AI in your own products. You should at least be thinking about that, too, in addition to experimenting with AI from other companies.
Recommended For You
Low-Power WANs offer an alternative to 5G for connecting a fast-growing array of basic devices and sensors that transmit small amounts of data.
An effective network visibility strategy requires understanding the technical, financial, political, and legal aspects impacting your network operations.
Emerging organizational structures for IT include placement of IT pros in user areas and departments forming their own "micro IT's."
Comparing a good and bad trace helps identify performance issues. Dynamic baselining can be used when you do not have a good trace to reference.
Combining commodity server platforms and FPGA-based SmartNICs will allow network applications to operate at hundreds of gigabits of throughput with support for millions of simultaneous flows.
SD-WAN implementations are on the rise thanks to the potential cost savings, increased network resiliency, and better application performance they deliver.