How AI Might Be Making Businesses Smarter
WHY YOU SHOULD CARE
Because maybe it’s time to embrace AI, not fear it.
By Laura Elizabeth
OZY and JPMorgan Chase & Co. have partnered to bring you an inside look at how entrepreneurs and their good business are helping the communities around them. Enjoy the rest of our special series here.
Mention artificial intelligence and it’s not long before you’re talking about mass job losses, robot revolutions, maybe even human enslavement. But a more realistic view of the technology is less crops of people-turned-batteries (thanks, Matrix movies), more handy programs for scheduling meetings. In fact, if AI continues on its current trajectory, it will likely be the very opposite of an enemy of the people or the catalyst to human obsolescence. Rather, AI could be a tool that frees up humans to focus on the tasks they do best.
It’s for this reason that it makes good business sense to be an early AI adopter — and that goes for small businesses too. Because while the world’s biggest companies might be busy investing billions in self-driving cars and the like, it only takes relatively simple AI to make a sophisticated business impact, and give people more time to be better at their jobs.
AI won’t be reaching out to an account holder or giving that personal touch. Humans still need to do that part.
Gwen Schlefer, Bonanza
Take Seattle-based online marketplace Bonanza, a fairly typical start-up (job roles include Director of Happiness). The company relies on satisfied buyers and sellers using the site again and again, and so they need a way to track if that’s happening. But, says Bonanza’s public relations manager Gwen Schlefer, “With over 40,000 sellers, 200 million items and hundreds of variables, the data sets are too large and complex to analyze using traditional statistical methods.” The solution: Bonanza built its own in-house AI software to identify changes in user behavior (such as a seller not logging in as regularly) and outcomes (like decreasing sales). Crucially, after the AI flags at-risk accounts, it is a human who contacts the users to offer help and boost satisfaction. “AI won’t be reaching out to an account holder or giving that personal touch,” Schlefer says. “Humans still need to do that part.”
A wise observation in business is that to be successful, you should only do what only you can do. So, say you’re a news editor, that means devoting your time to editing the news, not uploading stories or processing invoices. It’s seldom things turn out this way, however, particularly in small businesses, where employees will be managing administrative tasks like calendars and communications as well as more complex duties. But at Roger Wilco Agency, a small content marketing, branding and design firm in Dallas, AI provided an easy solution.
The agency’s founder, Mark Hopkins, has been using a virtual personal assistant — purchased from New York’s X.ai — for more than six months. He says “Amy” has been a huge time-saver. To organize a meeting, for example, all Hopkins needs to do is copy Amy in on an email to whomever he’s meeting with, and she’ll take responsibility for scheduling the date and time, swapping phone numbers and even adding real-time updates. “If you message the group to say you’re running late, Amy will push the meeting back and automatically cascade any subsequent meetings that might be affected,” Hopkins explains.
AI isn’t just useful for workflow, though. As Bonanza’s integration of AI into customer relationship management shows, it’s also becoming pivotal to business intelligence. The power here lies in technology’s ability to process more data than the human brain can handle. Competitive analysis tool Crayon, for example, not only tracks competitors in categories like price and brand messaging but also uses AI to identify competitors’ product gaps and weaknesses, helping develop a competitive business strategy.
It would be foolhardy to suggest AI is without flaws. First, a key facet of the technology is that it learns over time. “Early on, there were some configuration or user errors,” Hopkins admits, revealing that, at first, Amy couldn’t necessarily distinguish between a 9 a.m. meeting, the ninth floor or nine confirmed attendees. “But the accuracy has gone up over time,” he adds, an observation also held by Bonanza’s Schlefer, who notes: “We’re improving the AI by giving it more data, which helps it learn. We manually confirm the AI’s information, and tell the AI what happens with at-risk sellers, so it can learn by experience.”
The biggest problem with the technology is not the wait for it to learn, though. Rather, it is that even software engineers often cannot fully explain how their AI is learning. Programs like Amy are able to learn thanks to neural networks, or artificial brains with structures that mimic ours. And, with more complex AI, once these systems begin programming themselves (i.e., learning), it becomes near impossible to explain precisely how and why they reach their conclusions.
Why is this an issue? Because the mind of the program becomes a mysterious thing, and you don’t know what it might do next. That doesn’t necessarily mean your virtual office assistant will hack into the Pentagon and hold the human race to ransom. But it could mean, for example, that a self-driving car elects to mount a sidewalk for no reason a programmer can fathom.
Ultimately, though, we are only now scratching the surface of AI’s possibilities. Yet, as is always the case with new technologies, the pace of change is getting faster. Given AI is already a proven tool for maximizing business performance, companies would likely be wise to act now if they don’t want to be left behind.
- Laura Elizabeth, OZY Author Contact Laura Elizabeth