After a year of mini scandals involving articles created via generative AI, we’re finally starting to see AI applied to news in ways that don’t produce sloppy or embarrassing work.
Semafor Signals — one of several initiatives Microsoft cited this week in its sweeping announcement to support newsrooms and entrepreneurs applying AI to journalism — is a new format that leverages generative AI to create summaries of breaking news that reporters and editors can then curate, with heavy editing and adding their own analysis. (Members of The Media Copilot do consulting work for several companies, including Microsoft.)
By using AI to write the initial drafts, I expect the time to produce a Signals story, as opposed to a reporter scanning the news and writing their own copy, is significantly reduced. The tool also makes a point to look for articles about a topic in multiple languages, so reporters aren’t limited to only the languages they can read. Sure, Google Translate has been available for a long time, but its use in reporting is mostly limited to when the reporter knows exactly which articles to check. With the help of AI, you can effectively search and “read” multiple articles throughout the non-English-speaking web.
AI chatbots are only as good as your prompt. However, few people have the time or patience to test, iterate, and revise their prompts for the best results. Luckily, we’ve done that for you — download our free guide to the 100 best prompts for PR professionals or click the image below ?
Semafor Signals are clearly a big deal for Semafor. I can tell because co-founder Ben Smith told the Verge said so, explaining the company plans to redesign their entire web experience around Signals. Seemingly confirming this, a Signals story is leading the site right now (12:09 p.m. ET). While there doesn’t seem to be a dedicated section for Signals stories, a feed of them appears at the bottom of the introductory post, starting with the most recent.
As a use case of applied AI, Signals tick a lot of boxes. For starters, humans are in the loop. Generally, most of the egg-on-face incidents with generative content happened because human oversight was lacking. Also, it appears Signals stories, once written, go through a standard editorial process, meaning a section editor or subeditor is vetting the final copy.
How to Create Your Own Version of Signals
Signals are also nothing too special. Semafor was quite generous when it spoke to the Verge, revealing the workflow behind Signals: Reporters use an internal tool called MISO, which stands for multilingual insight search optimizer. Since this is a Microsoft tool, you can bet MISO is powered by OpenAI’s tech, which means you should be able to create your own version of Signals using ChatGPT or even Microsoft Copilot.
Leave a Reply