Enterprise leaders understand the power and potential of AI agents. They offer constant availability and can move autonomously, analyzing and collecting data to accomplish tasks quickly and efficiently. But realizing those gains requires winning over skeptical consumers. To better understand consumer sentiment around AI agents, Auth0 commissioned a global consumer survey with Statista. Here’s a look at what we learned. (All data in this article is derived from the survey.)
The findings, captured in our latest Customer Identity Trends Report, revealed that most consumers (70%) would rather communicate with humans over AI agents. Only 16% said they would prefer AI agents. To get the most from agentic AI, companies will have to build consumer confidence, and that starts with providing a secure, seamless experience that gives customers what they’re looking for.
Customers prefer humans for now
There is a disconnect between what some businesses want agentic AI to do and what many customers are ready for. We found that consumers are currently more willing to leverage AI agents when agents are used to perform monotonous and rules-based tasks, such as language translation, research, and writing assistance. Users reported less interest in using AI agents for project management, calendar and schedule management, and decision-making support.
Among those who would rather interact with humans, 64% said they felt a person would better understand their needs. Thirty-eight percent said that dealing with an AI agent can be frustrating, and 29% said outright that they simply do not trust them.
A human on the other end of a phone line, it seems, brings a certain level of confidence. If customers aren’t confident that agents will understand their needs or retrieve the correct information to help them, a customer support agent risks becoming a hindrance rather than an asset.
The solution to this challenge is clear: Companies must make user experience a cornerstone of their efforts to build consumer trust and comfort. The interaction between AI agents and customers needs to be smooth and personable. It should also be fast. More than half (55%) of respondents who said they prefer AI agents over humans said they do so because agents can resolve their requests more quickly. Additionally, 35% of customers who won't use AI agents raised concerns about the agents' reliability. These numbers are a sign that developers should focus on optimizing agent architecture and workflows to enhance agents' performance and output.
Security and privacy are top AI concerns for consumers
No conversation about agentive AI is complete without addressing security. Customers and developers have a shared goal: Both want agentic AI to be secure. Yet many consumers are unconvinced that AI agents will respect their security and privacy needs. The majority (60%) of respondents said they were either "concerned or very concerned" about AI's impact on their digital privacy and security. Just 9% felt it was of little or no concern.
Education can partially bridge this gap. Forty-four percent of those who don't use AI said, "I don't trust AI agents with my personal data." Explicitly disclosing what sensitive data the agent is accessing and how it’s used can be a powerful step toward building confidence.
Having humans in the loop may also help, but it risks introducing delays, human errors, and trading AI bias for human bias. On the positive side, this approach can train agents to better understand the nuances of a request and refine their actions. For their part, developers can take steps to build and deploy more secure AI applications.
Why identity security is key to navigating the AI era
When it comes to security and privacy, so much of it hinges on getting identity security right. Overpermissioned agents can result in data leaks and increase the potential for malicious activity. Imagine an AI agent tasked with suggesting products for your shopping list and accessing your medical records in response to an inquiry about an over-the-counter medication. Now, imagine that agent falling victim to a prompt injection attack from a threat actor.
The power of AI agents changes the attack surface that organizations must protect. It’s crucial to implement least privilege and enable dynamic updates to permissions to grant or revoke authorization when needed.
There’s still much work to be done, but there’s reason for optimism for enterprises. Among those surveyed who preferred AI agents over humans, 51% said they see AI technology as the future. For businesses, AI agents offer a path to efficiency, but in order to walk it, enterprise leaders will have to demonstrate to consumers they can offer a user experience that is fast, reliable, and — most critically — secure.
Download the full report to discover other crucial considerations for realizing the promise of AI agents without compromising customer trust.
