For months, a familiar storyline has plastered news sites as company after company announces that freshly adopted artificial intelligence tools have enabled it to slash hundreds or thousands of positions or refrain from filling open ones.
The implicit message is clear: AI has gotten good enough that firms need not bother with human talent anymore. But to experts who help companies make the most use out of AI, this narrative misses the value of what these tools offer.
Keep up with local business news and trends
Subscribe to the St. Louis Business newsletter to get the latest insights sent to your inbox every morning.
“I think for a long time, investors rewarded companies for hiring people. Hiring equaled growth and that was a proxy, and I would argue maybe even a lazy proxy,” says Andrew Sweet, who leads the AI solutions group at the Atlanta-based enterprise AI solutions firm AnswerRocket. (Sweet is based in St. Louis, where he says the company’s team is growing.) “We’ve exchanged one lazy proxy for another, and that is laying off people and blaming AI. I don’t think that is helpful to the market.”
Sweet says recent advancements in AI have given investors jitters over which companies will survive the disruption and which ones will go extinct. Businesses feel pressured to show they’re responding to these threats, but he argues cutting broad swaths of employees misses the real value that AI can unlock for workers and the firms they work for.
After all, simply getting employees to use AI doesn’t guarantee progress for either individual workers or a company overall. A growing number of companies (mostly tech firms) are moving to track and enforce the use of AI, but Sweet contends the more valuable metrics are the outcomes AI can power, which is mostly in the form of more efficient workflows that return time back to employees.
Sweet encourages the senior leaders he typically works with to reflect on the time they or their employees sink into repetitive tasks that could be automated to save time. But, he says, it’s equally important to direct workers to business problems or challenges that need tackling next once their schedules have been freed up.
“When you start to save time, that doesn’t mean that that’s an immediate opportunity to take out costs through employee reductions, but redeploy those employees doing jobs that humans were meant to do,” Sweet says. “If you’re creative as a senior leader in a company, there’s no shortage of problems to solve that require humans to solve it.”
Employees often need time set aside by their bosses to experiment with tools or formalized AI training, says Yongseok Shin, an economics professor at Washington University who researches macro economics, the labor market, and technology. Workers stretched thin who are pushed to use AI might think, “Who has the time? I mean, I’m just busy keeping up with all my work.”
“There’s tremendous value in demonstrating how it can be useful for people,” Shin says. “Unlike the traditional digital transformation, where firms just have this enterprise solution and just deploy it equally to everybody, AI doesn’t work like that.”
Michelle Hamilton agrees that this approach is critical for unlocking short-, mid-, and long-term wins from AI. Hamilton founded and led Spark AI Strategy, a local consulting firm that coaches companies on how to effectively make use of AI tools, before joining Sweet’s team at AnswerRocket this past March.
Just turning employees loose in the hopes that they’ll figure out how to make the best use of AI themselves isn’t helpful, she says.
“If you believed in your people enough to hire them, then you need to respect them enough to know that dumping a new tool on their desk and giving them a slideshow from IT is not enough,” Hamilton says, likening that approach to sending a person to a construction job site without knowing how to use the equipment they’d encounter there.
“Anyone can pick up a hammer, a saw, or a drill, but if you don’t know how to use those tools properly to build a building, you need to find someone to work with and get the right training,” she says. “Not just how to use it, but how to use it safely.”
The starting point for effectively using AI isn’t about the individual tools, she says, but rather what differentiates a company to its customers or users, be that its brand, reputation, or something else. Hamilton says these qualities should be used to train an AI bot that’s implemented internally to ensure it doesn’t dilute or deviate from what makes the firm special.
“It becomes very obvious when someone is just leaning into AI and saying, ‘Write this for me,’ without checking it for facts of hallucinations,” she says. “That is not a time saver. That is a reputation risk.”
Put another way, Hamilton says AI is like every individual employee getting “a very smart intern”—it can generate summaries of long reports or draft messages, but performs much better with context or background of what constitutes a strong response.
She and Sweet argue this isn’t something companies should put off as many likely already have workers using AI, whether their leaders are aware or not. Hamilton says this can make for problems if employees aren’t clear on how the technology applies to their role or if they’re dissatisfied with the options available to them.
“They will download something to their phone and more than likely upload sensitive data to an outside model because they are frustrated with either the training or the quality of what has been approved for them,” she says. “That is a governance and security risk you created by underinvesting in enablement.”
Hamilton says St. Louis firms have largely not been reactive when implementing AI, with “the ones that are doing it well taking a very measured approach.”
Compana, the parent company of pet brands including Bullymake, Doggie Dailies, Manna Pro, Oxbow and many others, is one example.
Before Hamilton helped the company apply Microsoft Copilot and ChatGPT, many Compana employees weren’t using the tools, says senior vice president of IT Pete Hogan. He says the training started broadly with prompt engineering and how to generally use the tools as a “digital assistant” before breaking out into individualized sessions for specific departments, such as legal, human resources, finance, supply chain, and even IT.
It’s allowed the company’s different departments to get more efficient or eliminate costs, Hogan says, using AI to kickstart new nutritional formulations, in the earliest phases of developing new packaging, or for focus groups.
“Not one of them is a game changer, but together, it’s people using these tools smarter, and they’re using them to offset things [where] we used to have to buy services [or] outsource,” he says. “Focus groups are very expensive, but if you prompt engineer properly, you can give the tools the same information you would feed a focus group and get very similar outputs that a focus group would give you.”
The company now holds regular meetings to identify new places where AI can potentially make a difference, Hogan says. Compana’s legal department has already implemented a broader tool that can review contracts for key terms or clauses.
“It used to take them hours per contract, now it takes them a fraction of the time,” Hogan says. “They focus on the areas that are unique to that deal, so it’s an incredible time save, which translates to cost.”
Demonstrating and explaining these kinds of outcomes has been essential to getting more employees on board, he says. AI helps boost the company’s output without needing to grow headcount “in a cost challenged environment that we’re all in today.”
“I’ve run into people [who’ve] actually stopped me in the hallway and said, ‘I can’t believe I didn’t use this sooner,’” Hogan says. “What we have is people that work really hard, that put in a lot of hours. We’re trying to help them understand this can ease your job.”
Hogan and Hamilton acknowledge that some people will be anxious or resistant to using AI, which is why robust training matters.
Hamilton says she will sometimes train that group of employees on the tools first before the rest of a company, usually demonstrating their usefulness through anything but core work functions. She describes one session where she walked people through cooking a recipe or planning a trip to Italy as a way to loosen them up.
“My favorite part is that people laughed. They had fun. Nobody raised their hand and said, ‘This is horrible. You’re just trying to replace my job,’” she says. “They had a great time planning this amazing trip to Italy. And then when you connect it back to their actual work, the resistance drops because they’ve already seen what it can do.”
This perception is important, Shin says. He recently supervised student-led research into how workers feel about AI and what drives their willingness to adopt it at work. “People who have high expectations of what AI can do for them actually tend to use more and feel like they get more benefit,” he says.
That aligns with Hamilton’s experience. When she returns for a broader AI training, she often finds that the employees who were once-resistant or skeptical can be the best teachers for their colleagues. After all, they themselves have already experienced what the tools can do.
