As artificial intelligence tools rapidly integrate into workflows, a growing number of employees are adopting them independently, prompting business leaders to address not whether AI is being used inside their organizations, but how to manage its use.
“We’re past the binary decision of ‘should I or should I not’ use AI,” said Purushottam Papatla, a business professor at the University of Wisconsin-Milwaukee who served as the co-chair of the university’s AI task force.
“Now the questions are more about how we come up to speed,” said Papatla. “How do we ensure that everyone in the organization comes along, and that it doesn’t create friction? How do we ensure that we acculturate our employees to using AI as a companion, rather than seeing it as a substitute for work or a competitor for their jobs? And so on.”
[gallery columns="2" size="full" ids="615044,615043"]
Informal adoption outpaces policy
Although AI has long been in development, the recent wave of publicly accessible generative tools like ChatGPT and Claude AI caught many organizations off guard and often spread outside of formal channels. Recent studies show that fewer than 10% of companies have official AI policies in place, despite more widespread usage, with a higher percentage of employees reporting using AI.
At this stage, executives should assume AI is already in use within their organizations, business school professors say.
“At every company there will be some who will be early adopters and innovators who go out and start using it on their own because they’re excited and they’re interested,” Papatla said. “I think companies need to be aware of the fact that AI is probably being installed already by at least a percent of their employees.”
Gene Wright, assistant professor at MSOE’s Rader School of Business and an executive with 40 years of corporate experience, echoed that sentiment. He added that employees often underreport AI usage.
“It’s common for employees to say they’re not using AI, even when they are,” Wright said, “because a lot of people still feel like it’s cheating, or it’ll somehow result in them getting replaced.”
This creates a hidden layer of risk, especially when the technology is used on sensitive business data. Generative AI models are trained on input, and many store data in ways that aren’t always clear to the end user.
“If somebody is using it to help them write a better customer service memo, or to help write better marketing copy, those are really good things, I think,” Wright said. “If you say, ‘Here’s my last 12 years’ worth of profit trends. Help me analyze that.’ In that instance, they’re giving away company data.”
Building an AI policy framework
According to both experts, the first step for any organization to develop a policy on AI is to map current AI usage. From there, a baseline policy that addresses key concerns – such as data protection, appropriate use cases and the approval of specific tools – can be established.
“The first step would be to start with an assessment of how extensive the use of AI already is among employees, and based on that, assess the urgency of coming up with policies,” said Papatla.
Despite the risks, both Papatla and Wright believe organizations should not discourage early adopters outright and said that working collaboratively with employees, especially early adopters, to shape guidelines could help ensure the policy is respected.
“I think when people are coming up with a policy, you should try to get people who are embracing it to help inform the policy,” Wright said.
“You really need to develop your policies somewhat organically, yes, but there are best practices for AI policies and consultants to help with that,” Papatla added.
Both experts also emphasized that policies must address the sensitivity of data. Not all AI usage carries the same level of risk, but some sectors, such as finance, health care and education, require more caution.
“The more sensitive the data is, the more urgent it is for the companies to come up with policies to ensure that employees who are using AI anyway are aware of that,” Papatla said.
Wright believes that encouraging thoughtful AI use can improve collaboration and idea-sharing – but warns that overreliance can have a downside.
“If you encourage people to use AI, it’ll undoubtedly boost your productivity, and I think you’ll also get people sharing with one another and learning from one another,” Wright said. “If people stop thinking critically and just accept what the tool says, that’s a problem. … ‘Brainrot’ is real.”
