Inc. Magazine
Relying on AI at work might seem smart, but it could cost more than you think. While AI boosts productivity, it also brings serious legal and ethical risks. From misinformation and liability to rising costs and public distrust, experts warn that without human oversight, AI could land businesses in real trouble.
This Expert Says You May Be Overlooking Serious Legal Issues From Using AI at Your Company
Author: Kit Eaton
AI is one of those world-changing technology waves that washes in fast and then is seemingly everywhere—in tools we use for fun, in places where AI seems a surprising addition (fancy an intelligent toothbrush, anyone?) and definitely in the office. For smaller businesses, adopting AI seems like a no-brainer. It can add capabilities to your team to make up for skill gaps, it can boost worker efficiency, and agent AI may soon be able to take on some of the boring office tasks that eat up otherwise useful work hours. But when a transformational technology like AI arrives, it’s easy to skip over the part where it also brings risks as well as benefits. MJ Jiang, chief strategy officer at New York-based small business lending platform Credibly, is savvy to these issues, however, and she’s issuing a warning: don’t overlook both the costs and the ethical considerations of using AI tools at work. And ask yourself the question: who will get sued when a gen AI-tool makes a mistake?
Some people see AI as a powerful way to cut costs. And one of the key ways that some managers are thinking of doing this is by outright replacing some workers with AI tools. On paper, dispassionately, it makes sense: AI’s don’t need health insurance, they don’t go on strike, and they can work 24-7-365 without getting tired. But to Jiang, this is all the wrong way to think about things. In an email interview with Inc., Jiang agreed that some firms will use AI “with the explicit goal of keeping humans in the loop but making them more productive and efficient” but even so there will be others that “dive into AI with the explicit goal of removing humans.”
How effective AI is for these businesses depends on how you choose to measure it, of course, and also on whether you’re trying to replace people or augment their skills. The “bottom line,” Jiang says, is that “because you cannot eliminate, only mitigate, hallucinations”—inaccurate information presented as fact—AI tools can introduce risks if you use it in a business decision. And “because GenAI cannot explain to you how it came up with the output, human governance will be essential in businesses where the use cases are of higher risk to the business.” It’s even possible that “human interaction will actually be valued at a premium, because customers will demand talking to a real human being,” Jiang says.
That might sound like an endorsement of using AI to drive efficiencies at work, as long as there’s a reliable human hand on the steering wheel, but Jiang added a cautionary note. She thinks the fact that AI is “opening up new liabilities has not been well understood yet by businesses.” This is because our “institutional knowledge and regulatory framework are designed for understanding liability when human actors are involved,” and when regulations do evolve, they do so slowly unlike “AI which is exponential.” There’s almost a different “kind of math that companies need to understand,” Jiang said, adding “it’s not just looking at costs of AI today versus cost of wages, it’s the future cost of GenAI given concentration of pricing power in the hands of few tech companies, plus added liability, plus costs of monitoring and governance.” In other words, even if you think AI tools will save you money, when viewed in the long term, and baking in the risk factors, this is difficult situational modeling to do, which might be why some smaller companies are risking tricky or expensive legal issues by not building contingencies for AI-related issues into their long term business plans.
Another thing Jiang thinks companies adopting AI tools should do is be conscious of the disparity between what some people say is the trustworthiness and risk factor of AI and what the general public thinks. She quoted a recent Salesforce survey which shows 43 percent of the public don’t trust companies to use it ethically, and 89 percent want to know if they’re communicating with an AI or a person, and, perhaps most importantly, 80 percent want humans to validate the AI inputs. No matter if your company has done its due diligence and decided AI’s risks are worth it, the public may not trust you if you assure them that your products that include AI—or perhaps just your AI-assisted business operations—are trustworthy.
Why should you care about this? For one main reason: Credibly’s and Jiang’s insights show removing people from the loop and subbing in an AI tool not only can drive up costs of your business (when you were hoping they’d fall), but it also raises serious accountability issues when AI makes mistakes. And that latter point, sadly, is inevitable.
Credits: TCA, LLC.