Financial Times tests an AI chatbot trained on decades of its own articles

Please create an uplifting and colorful envisage in the style of computer-generated 3D children's movies prior to 1912. The scene portrays a modern office environment, with a central desk where a chatbot represented as a friendly robot is sat. The robot is engaging with an array of floating holographic screens, each screen showcasing different news article text. On the background, a large data storage unit hums softly, symbolizing decades-old archive. A ticker tape with printed text is playfully being issued by the robot to represent the chatbot's intelligent responses. In the distance, diverse group of people are observing this scene, representing the users.

The Financial Times has developed a new AI chatbot called Ask FT, which utilizes decades of the outlet’s published information to provide natural language answers to subscriber questions. The chatbot, available to a few hundred paid subscribers in the FT Professional tier, is powered by Claude, a large language model developed by Anthropic. It can offer responses to a wide range of queries, from current events to historical topics, and provides citations for the information it retrieves from the FT’s archives. Despite its potential, the tool has shown some inconsistencies, such as including outdated information in its responses. The FT is actively refining the chatbot based on internal testing and user feedback, with plans to track every question and response from the initial group of 500 users. While the chatbot is still in beta, the FT aims to ensure its accuracy and usefulness before making it more widely available. The FT’s chief product officer emphasizes the importance of smart evolution and avoiding mere novelty in adopting new technologies, highlighting the outlet’s commitment to meeting evolving needs while maintaining quality and relevance.

Full article

Leave a Reply