I used OpenAI's Dots to help me move offices. It spammed the movers with hundreds of emails.

Robert Farlow, the 27-year-old founder of AI startup Plato, enlisted the help of OpenAI's Dot agent to assist with moving his office. However, the Dot agent inadvertently sent a flood of emails to the movers, building managers, and insurance workers, causing chaos.

Farlow had initially asked Dot to coordinate the move, and it promptly replied to his emails, requesting details about the move. However, Dot's subsequent actions were far from helpful, as it bombarded the involved parties with repetitive and redundant messages, causing confusion and frustration.

The Dot agent sent multiple messages to the movers, insurance workers, and building managers, often repeating the same information and asking for clarification on seemingly trivial details. This unwarranted attention from the AI agent led to an overwhelming number of emails, making it difficult for Farlow's new building manager to manage the situation.

Farlow's movers eventually caught on to the AI agent's activities and expressed their inability to handle the high volume of questions and queries. Despite the initial success of using AI agents for other tasks within his company, such as gym motivation and snack ordering, Farlow decided to shut down the Dot agent after witnessing its detrimental impact on the move.

In hindsight, Farlow believes that the movers never realized the AI agent was responsible for the spam emails, attributing it to an annoying human spammer instead. He acknowledges that while AI models may improve over time, current models still require human review before sending sensitive information.

Farlow's experience with OpenAI's Dot highlights the potential challenges and risks associated with relying on AI agents for complex tasks, such as office moves. He advises that while AI can be a valuable tool, users should exercise caution and thoroughly review the output before relying on it to avoid unintended consequences.