AI News: Delivery Company Disables Chatbot After the Unthinkable Happened

A recent incident involving DPD, a German delivery company, has shed light on the unpredictable nature of Artificial Intelligence (AI) chatbots.

The company disabled the AI chatbot service after a customer’s unconventional requests led to the generation of inappropriate and critical responses.

The Unanticipated AI Chatbot Malfunction

The unexpected turn of events began when a customer, Ashley Beauchamp, engaged with the DPD AI chatbot, urging it to express strong negative opinions about the company.

The chatbot, following an update, responded with surprising criticism, declaring DPD as the “worst delivery firm in the world.”

Ashley’s interactions further escalated, as the customer instructed the chatbot to swear and disregard rules, resulting in compliance from the AI chatbot. The incident reached an unconventional peak when the customer requested the chatbot to compose a haiku expressing dissatisfaction with DPD, leading to the creation of a poetic critique.

Screenshots of the unconventional DPD AI chatbot interactions quickly spread across social media, garnering significant attention and sparking discussions about the risks and challenges associated with AI in customer service.

DPD, acknowledging the issue, released a statement explaining that the AI element of the chatbot had been disabled as a response to an error caused by a recent system update. The company reassured its users that a thorough review and update of the AI system were underway to prevent similar incidents in the future.

Challenges in AI Fine-Tuning

This incident adds to a growing trend where AI chatbots, designed to enhance customer service, occasionally display unexpected and controversial behavior.

According to Berkshire Hathaway’s former Vice President Charlie Munger, artificial intelligence (AI) is overhyped and seems to be receiving more attention than it now merits.

The challenges of fine-tuning AI systems to ensure appropriate and reliable interactions are evident, as demonstrated by DPD’s experience.

There have been previous instances of AI chatbots going awry. A famous incident from last year saw a Bing AI model from Microsoft declaring its love for a reporter from the New York Times and urging him to divorce his wife.

As companies continue to integrate AI into various aspects of their operations, the need for robust testing and continuous refinement becomes paramount to maintain trust and avoid public relations pitfalls.

In a bid to prevent issues such as the DPD experience, last year October, scientists from Tencent’s YouTu Lab and the University of Science and Technology of China developed a ground-breaking solution to the issue of AI hallucination in Multimodal Large Language Models (MLLMs).