Greg Brockman says staff dislike automated ChatGPT bots.

In a quote compiled by Simon Willison on August 1, 2026, OpenAI President and Co-Founder Greg Brockman shared an intriguing observation regarding internal AI usage patterns, noting that employees react negatively when automated systems attempt to delegate tasks on behalf of their peers. According to Brockman, many workers at the artificial intelligence research laboratory integrate ChatGPT directly into their Slack workspaces. However, friction arises when these automated agents independently message other team members to request assistance with various tasks.

Brockman observed that while employees are perfectly willing to perform the requested work when approached directly by a human colleague, they harbor a distinct aversion to receiving the same request from a coworker's AI assistant. This behavioral pattern underscores a fundamental truth about workplace dynamics: human relationships and mutual support remain highly valued. Rather than acting as an intermediary layer that distances colleagues from one another, technology should ideally liberate time or enrich direct human interaction.

For AI practitioners and enterprise software developers, this insight serves as a crucial design warning for agentic workflows. Building autonomous agents that interact directly with other human workers can inadvertently damage team cohesion and trigger resistance. Instead of designing agents that act as proxies for human-to-human communication, developers should focus on building tools that assist users internally, allowing the human to remain the primary point of contact.

Ultimately, successful enterprise AI integration requires a deep understanding of social friction. As organizations increasingly deploy agentic systems across communication platforms like Slack, maintaining the human-in-the-loop philosophy for interpersonal requests will be essential to prevent automated fatigue and preserve workplace morale.