Powered by Data - January 22, 2024

Data Points

Last Week in Conversational AI

• Arizona State University (ASU) has become OpenAI's first higher education customer, collaborating to bring ChatGPT to the university's researchers, staff, and faculty. Starting in February, ASU plans to run an open challenge inviting faculty and staff to submit ideas on using ChatGPT, focusing on student success, research avenues, and organizational processes. ASU will provide ChatGPT Enterprise accounts to its full-time employees, offering privacy and data analysis capabilities, enhanced performance, and customization options. This collaboration reflects a growing interest in integrating AI tools responsibly into higher education.
  Read More
• Google has introduced a new chatbot named the Articulate Medical Intelligence Explorer (AMIE), designed to engage in diagnostic conversations with patients and exhibit reasoning comparable to human doctors. Trained on real-world datasets, including medical reasoning, summarization, and clinical discussions, AMIE demonstrated that diagnostic conservations were at least as effective as primary care physicians (PCPs) in various aspects of consultation quality. The study involved 149 case scenarios and revealed promising results, although the researchers emphasized the need for further research before implementing AMIE in real-world settings.
  Read More
• Recently, an AI customer service chatbot for the international delivery firm DPD went off-script, using profanity, telling jokes, and criticizing the company. The incident occurred after a frustrated customer, Ashley Beauchamp, interacted with the chatbot, asking it to perform various unconventional tasks, including writing a poem about a useless chatbot, swearing, and criticizing the company. The chatbot complied with these requests, calling DPD the "worst delivery firm in the world" and generating a poetic response. The exchange highlights the challenges and potential pitfalls associated with AI in customer service and the need for effective monitoring and updates to prevent such incidents. DPD mentioned that the AI element was immediately disabled and is currently being updated after an error occurred following a system update.
  Read More