When the The Financial Times (FT) created its AccelerateAI team earlier this year, trying to navigate through the noise and explore both opportunities and threats.
“People are tired of talking about whether AI is going to kill us or save humanity,” team leader Liz Lohn, FT’s chief product officer, said at a press conference. The Future of Media Technologies conference last Thursday.
From the beginning, the AccelerateAI team decided that they would never jeopardize readers’ trust. They were willing to experiment, but also to shut down if something went wrong.
But generative AI isn’t the only solution. According to Lohn, the technology can be used for two distinct solutions: story creation and discovery, and story dissemination.
Content creation
The first lesson the team learned was humility, as it became clear that the value of AI-produced content depends solely on the quality of the tool itself. Can it produce summaries that are both informative and engaging? Or translate them to incentivize and create sufficiently relevant contributions?
The team spent six weeks training a large language model (LLM) and developing a prompt for a summary. When they saw the disappointing result, the newsletter team didn’t buy in.
“It was the best kind of failure, we learned a lot from it,” Lohn said.
Discovering new computer science
The next experiment involved using AI to extract information from structured data sets and generate new story leads. They decided to look at MPs’ income and expenses.
The problem with generative AI, however, is that it invents things (it hallucinates).
“Hallucinations are a bug in media and a feature of LLMs,” Lohn said, adding that generative AI will work best as an “inspire me” tool rather than a content creator.
“As soon as it works, no one will call it AI anymore,” she joked.
Sam Gould, head of AI at FT Strategies, who joined Lohn on stage, sees three clear shifts in user behavior.
- products become multimodal
- audiences engage in conversational experiences
- and automation tools become agents
Multimodal products
Video is becoming an increasingly important source of online news, especially among young people. AI-powered tools can now create text, images, audio, and video, allowing publishers to experiment with different formats and features, as well as internal processes (transcription, translations, creative analysis, data journalism), all of which involve processing multiple types of unstructured data. Gould cited a practical example of analyzing footage from live reporting in the field, where AI was able to instantly translate a road sign written in another language.
Want a chatbot?
While most people shudder at the thought of a long “conversation” with a bot, it can be a valuable tool for your audience to explore content. Much like short-form videos, chatbots are proving particularly popular with younger audiences.
AI Agent
Gould concluded by saying that AI tools are being tasked with increasingly complex tasks, such as writing code or creating new applications with minimal human intervention.
So, love it or hate it, AI is changing the publishing industry and we need to prepare to make the most of the opportunities while mitigating the risks.
Free daily newsletter
If you like our news and feature articles, you can register to receive our free daily newsletter (Monday to Friday) electronic newsletter (compatible with mobile devices).