AI Agents Need Trust and Regulation to Succeed

Bloomberg Opinion — The buzzword "agents" dominated discussions at Taiwan’s Computex, Asia’s largest tech conference, as executives like Nvidia CEO Jensen Huang touted them as future "digital employees." These AI tools, which go beyond chatbots to execute tasks autonomously, are being pitched as the next paradigm shift, with demos showcasing capabilities from creating marketing presentations to managing household tasks.

The Rush for Agentic AI

Tech companies, including Google, OpenAI, and startups like Sakana AI and Manus, are racing to monetize AI agents, targeting enterprises for profitability. However, this push comes amid declining public trust in AI systems, with KPMG reporting a drop from 63% in 2022 to 56% in 2024.

Challenges to Adoption

1. Trust and Accountability: Without robust governance and adherence to international standards, AI agents risk failing to gain widespread acceptance. Policymakers, particularly outside Europe, have been slow to implement regulations, leaving gaps in oversight.
2. Technical Barriers: Granting AI agents access to password-protected or sensitive data remains a challenge. The transition from chatbots to reasoning models requires significantly more processing power, as noted by IDC.
3. User Skepticism: Demos, like Manus AI, reveal limitations in handling simple tasks (e.g., restaurant reservations) and raise concerns about sharing sensitive information with autonomous bots.

The Path Forward

For AI agents to succeed, the industry must prioritize:

• Building trust through transparency and accountability.
• Addressing technical hurdles, including security and scalability.
• Implementing regulatory guardrails to mitigate risks.

As Catherine Thorbecke argues, this next phase of AI must unfold as a marathon, not a sprint, to avoid eroding trust and ensure sustainable adoption.

More stories like this are available on bloomberg.com/opinion.