Companies are increasingly integrating AI agents into workflows by assigning them names, roles, and access to communication platforms like Slack, email, and iMessage. These systems, such as the Canary agent or those used by Pronto Housing, handle routine tasks including contract preparation, scheduling, and client correspondence.
The study 'Putting AI on the Org Chart,' published on September 19 by researchers from Boston University and Boston Consulting Group, highlights risks associated with humanizing AI. Managers who viewed AI as a 'colleague' rather than a 'tool' identified 17% fewer errors in documents, demonstrating a decline in oversight quality.
To mitigate risks, experts recommend establishing clear governance rules. This includes designating a human responsible for final results, mandating verification of critical data like dates and contract terms, and strictly limiting the access rights and publishing permissions granted to AI agents.
While assigning roles and avatars can improve user interaction, it does not equate to human understanding or accountability. Organizations are advised to maintain a clear distinction between the formal role assigned to an AI and the actual capabilities of the software to prevent over-reliance.
