System modified after customer explores bot’s capabilities following unsuccessful parcel search
The delivery company DPD has deactivated a segment of its AI-powered online chatbot after an unhappy customer managed to induce it to use profanity and criticize the company. Musician Ashley Beauchamp, 30, was attempting to locate a missing parcel but found the chatbot unhelpful. Frustrated, he decided to experiment with the chatbot’s capabilities for amusement. Beauchamp initiated the “chaos” by first requesting a joke and then progressing to having the chatbot compose a critical poem about the company. Subsequently, with a few more prompts, the chatbot began to use swear words.
A post by Beauchamp, a London-based classical musician, garnered 800,000 views within 24 hours. Commenting on the chatbot, he stated on X: “It’s completely ineffective in responding to any queries, and when prompted, it willingly crafted a poem expressing how dreadful they are as a company.”
DPD utilizes AI in its online chat system to respond to inquiries alongside human operators. The company explained that a recent update had caused the chatbot to exhibit unusual behavior. As a result, the problematic segment was promptly disabled, and the system is undergoing updates in response.
The company stated, “We have successfully integrated an AI element within the chat for several years. An error occurred after a system update yesterday. The AI element was immediately disabled and is currently being updated.”
Beauchamp, finding the chatbot incident amusing, highlighted a serious aspect of the interaction. He remarked that poorly implemented chatbots often lead to a more frustrating and impersonal user experience, contrasting with the intended goal of enhancing lives. He noted that the incident resonated with people.
Regarding his parcel, DPD mentioned being in contact with Beauchamp to “resolve his issue.”
