A Reddit post detailing a manager's growing dependence on Anthropic's Claude has drawn widespread attention, as the employee described a workplace where the AI's outputs overruled staff expertise. The story, posted on Tuesday, triggered a wave of similar accounts and renewed concerns about automation bias eroding human judgment in corporate decisions.
'We need to appease Claude'
The employee, writing under the title "My boss has AI psychosis and we're f****d," said meetings increasingly centred on conversations the manager had with the chatbot. Targets drawn from those exchanges were presented as non-negotiable. "Go to meeting. Boss shares the conversation he had with Claude and stone cold serious tells us we need to appease Claude and meet 'his' projections," the employee wrote.
When those projections missed client realities, the manager did not question the AI. Instead, the team was blamed for failing to deliver. The manager's deepening attachment became a strain on the whole group. "I'm watching him get deeper and deeper into the hole and it's depressing to watch," the post read. "This can't be good for his mental health, it's not good for any of us."
The employee's account references Anthropic's Claude, a large language model the company markets as a reliable assistant - not a leadership replacement.
Community reaction: 'I don't want everything to be AI'
The thread filled with accounts of executives treating AI outputs as objective truth. "I don't want everything to be AI. I want real people. I don't see how so many people bought into this," one user wrote. Another commenter put it bluntly: "I swear we're all just finding out how many people wanted nothing more than some kind of entity which gives them all the answers and does all the thinking for them."
Others described concrete business damage. A commenter said their CEO now runs virtually every decision through an AI chatbot before approval, stalling product launches for months because AI-generated criticism makes leadership reluctant to sign off. Another recalled walking into work to find the CEO had distributed lengthy AI assessments criticising multiple departments. That scenario underscores why leadership training focused on AI for Executives & Strategy is drawing more attention from boards.
The erosion of human judgment
Experts have long warned that large language models can produce convincing but incorrect conclusions, misinterpret context, and reinforce biases. Human oversight remains essential, especially when decisions affect employees, budgets, or product strategy. Yet the Reddit thread illustrated a workplace where oversight had vanished.
The episode reinforces a critical message for middle managers: integrating AI effectively means knowing when not to rely on it. Structured programmes in AI for Management aim to build exactly that discernment, equipping leaders to calibrate when to trust machine outputs and when to push back.
Why this matters for management
Managers who delegate authority to AI without clear protocols risk two failures. First, they make poor decisions based on data that lacks operational context. Second, they demoralize teams whose expertise is overridden by a system that cannot grasp ground truth. The Reddit post is a warning that the cost of treating AI as a decision-maker, rather than a decision-support tool, shows up first in employee trust - and soon after, in business results.
Your membership also unlocks:
