The Financial Times has launched an AI-powered chatbot called Ask FT. By using Anthropic's Claude 3 model, this tool allows FT Professional subscribers to access information from the publication's extensive archive through natural language conversations instead of traditional search queries.
What sets Ask FT apart from general AI chatbots is its exclusive focus on the FT's own content. Instead of drawing from the entire internet, the chatbot sources answers directly from the newspaper's published articles, complete with citations and publication dates. This kind of walled-garden approach attempts to address two shortcomings of AI: Information reliability and copyright compliance.
Results have been mixed. According to Techiexpert, when asked about Microsoft's AI leadership, Ask FT correctly identified Mustafa Suleyman's role and cited recent FT articles as sources. Of course, this isn’t exactly proprietary information, and a simple Google search would turn up the same information. Ask FT also incorrectly listed Nikki Haley as a candidate in the 2024 US presidential election even though she withdrew from the race early on. This shows that even a carefully curated knowledge base isn’t enough to prevent misleading answers.
Ask FT represents an attempt to add value to premium subscriptions while also making use of the FT’s extensive back catalog of content, but whether the reality will match the hype remains to be seen.
To see how other news organizations are incorporating AI into their workflow, check out this piece about The Washington Post or this one about the Associated Press and the BBC.
About the author,
José Mauricio Duque
