Dive Brief:
- Employees are less comfortable and confident with AI use than their managers, and the power dynamic between AI decision-makers and those instructed to use the technology will likely continue that trend, a TeamViewer report released Wednesday found. The remote access software company worked with Sapio Research to survey 4,200 managers and employees globally this spring.
- Although three-quarters of respondents — both employees and managers — said they use AI at least once a day, 61% said they’d prefer it if AI had no independent action abilities. More than half of respondents verify AI outputs often or always before using them, and about the same proportion said they don’t always know when to trust AI and when to verify it, according to the report.
- Organizations must work toward trusted autonomy for their AI systems instead of blanketed automation, said Oliver Steil, CEO at TeamViewer, in a press release. “AI is already changing how people work, but the next phase will be defined by trust,” he said.
Dive Insight:
Although managers and employees alike are getting more comfortable using AI in their workflows, those leading adoption efforts generally feel better than frontline workers about AI and are more willing to delegate work to it. Employees reported that they feel greater personal risk from autonomous systems because of their lack of control over how and when AI is deployed.
Software has traditionally been designed around what people tell it to do, but AI systems and agents change that model, Mei Dent, chief product and technology officer at TeamViewer, said in a press release.
Insecurities around AI outputs aren’t causing workers to fully avoid the technology, but it might be limiting what they’ll use it for. AI is most frequently used once or twice a day, the survey found, and usually for tasks that help a user understand information, generate a starting point or complete a routine task — outputs that can be reviewed before they’re used.
Only 11% of respondents said they’re comfortable with AI working autonomously, or taking action that they haven’t reviewed and approved.
Despite lagging trust in AI systems, enterprises are forging ahead on giving them autonomy. One-quarter of business and technology leaders said they have complete trust in their AI systems, a June survey published by Kyndryl found, but more than four-fifths said they expect autonomous AI agents to make decisions with material business impact within the next year.
Organizations struggling with trust often lack governance and accountability frameworks needed for AI outputs, Steil said. Respondents said they worry that AI systems could expose sensitive information, that it might not understand the context of the information to create the right outputs and that AI could act beyond its intended scope. About one-fifth of respondents said they’re concerned they will be held responsible for AI’s mistakes.
“[Organizations] will succeed by building systems that operate safely, transparently and accountably within clearly defined guardrails,” Steil said. “The real opportunity is to use AI and automation to prevent disruption, improve digital experiences and free people to focus on higher-value work.”
Organizations must give their staff the confidence that autonomous systems are secure, explainable, reversible and governed by people, Steil said.
A good starting point is establishing which actions are low-risk and pre-approved to be completed autonomously and which ones require human review. Employees should be involved in the process of determining autonomy to understand the authority a system has and where responsibility for errors lies.
Organizations should also work toward ensuring that the risks of autonomous systems are distributed fairly across different workplace groups, rather than lying with one team or individual.
“Autonomy will scale when organizations can prove it works in specific, trusted use cases, then expand it with confidence,” Dent said.
