This voice is automatically generated. Please let us know if you have any feedback.
Diving overview:
- A TeamViewer report released Wednesday found that employees are less familiar with and confident in using AI than their managers, and the dynamic between AI decision-makers and those directed to use the technology is likely to remain that way. The remote access software company, in collaboration with Sapio Research, surveyed 4,200 managers and employees around the world this spring.
- Three-quarters of respondents, including both employees and managers, said they would use AI at least once a day, while 61% said they would use AI if it did not have the ability to act independently. The report found that more than half of respondents often or always verify AI output before using it, and a similar proportion said they don’t always know when to trust AI and when to verify it.
- TeamViewer CEO Oliver Still said in a press release that organizations need to work toward trusted autonomy for AI systems, rather than comprehensive automation. “AI is already changing the way people work, but the next step will be determined by trust,” he said.
Dive Insight:
Managers and employees alike are becoming accustomed to using AI in their workflows, but those leading adoption efforts generally feel better about AI and are more willing to delegate work to it than front-line employees. Employees reported feeling greater personal risk from autonomous systems because they have no control over when and how AI is deployed.
While software has traditionally been designed around human commands, AI systems and agents change that model, May Dent, TeamViewer’s chief product and technology officer, said in a press release.
While concerns about the output of AI are not causing workers to avoid the technology entirely, they may be limiting the purposes for which AI can be used. The study found that AI is most often used once or twice a day, usually for tasks that help users understand information, generate starting points, or complete routine tasks – outputs that can be reviewed before use.
Only 11% of respondents said they would be comfortable with AI operating autonomously or taking actions that have not been reviewed or approved.
Despite the lag in trust in AI systems, companies are working to give them autonomy. According to a June survey published by Kyndryl, a quarter of business and technology leaders say they fully trust their AI systems, while more than four-fifths say they expect autonomous AI agents to make decisions that have a significant impact on their business within the next year.
Organizations that struggle with trust often lack the governance and accountability frameworks needed for AI artifacts, Steil says. Respondents said they were concerned that AI systems may expose sensitive information, that AI systems may not be able to understand the context of information and create appropriate output, and that AI may operate beyond its intended scope. About one-fifth of respondents said they were concerned that they would be responsible for AI mistakes.
“[Organizations] “We succeed by building systems that operate securely, transparently, and accountable within clearly defined guardrails. The real opportunity lies in using AI and automation to prevent disruption, improve the digital experience, and free up people to focus on higher-value work,” Steil said.
Organizations must give staff confidence that autonomous systems are safe, accountable, reversible and controlled by humans, Steil said.
A good starting point is to establish which actions are low-risk and pre-approved to be completed autonomously, and which actions require human review. Employees need to be included in the process of determining their autonomy so that they understand what rights the system has and who is responsible for errors.
Organizations should also strive to ensure that the risks of autonomous systems are distributed equitably across different work groups, rather than relying on one team or individual.
“If an organization can prove that autonomy works for specific, trusted use cases, then autonomy will scale and can be scaled with confidence,” Dent said.
