AI-generated content could mean many things today, including personalized news feeds catered to readers’ interests and written articles gleaned from a data set. With so much information available 24/7 at our fingertips, it makes sense to have AI step in and help curate some of the news we consume. “It just doesn’t make sense for humans to be going in and generating and putting together certain types of content because there’s no creativity—it’s just mind-numbing, and AI systems are good at the mind-numbing thing,” Ronald Schmelzer, a managing parter and principal analyst at Cognilytica, told Lifewire over the phone.
Microsoft Start
Microsoft introduced Microsoft Start in September as a “personalized news feed and collection of informational content [that] provides news from premium publishers, timely updates tailored to your interests, and is available when and where you want it.” The platform includes news and media channels from over 1,000 publishers, combined with AI and machine learning algorithms to sort through which news is presented to users. It also personalizes content based on interests and how you engage with it. Microsoft Start is rolling out as a mobile app, a standalone website, and on the new Microsoft Edge Tab page. Microsoft Start comes after the tech giant invested $1 billion into open APIs in 2019—notably, a type of AI called GPT3. “GPT3 is a big AI tech text generator,” Schmelzer said. “You can give it just a small little bit of prompt text, and it can come back with literally paragraphs of text. Microsoft put a billion dollars into it, so you better believe they’re going to try to find every single possible way that they can squeeze value out of this thing.”
Choosing the News You See
Experts say Microsoft is tapping into a very particular type of AI pattern in its Microsoft Start platform. “It looks like Microsoft is really getting into this hyper-personalization pattern where they’re starting to tailor news towards that individual,” Kathleen Walch, a managing partner and principal analyst and at Cognilytica, told Lifewire in a phone interview. “With the help of artificial intelligence, you’re able to personalize your news because you don’t have print editions that people are reading—they’re reading online. And so you’re able to tailor that news to their liking and show them things that they like and understand their dislikes, and then don’t show them articles based on that.” This type of personalized news feed isn’t anything new, but it has seen expansion over the years. For example, Facebook launched a “related articles” function in 2017 to expose users to stories with different views than their own. Google News also hasrelied on AI since 2018 to put together information in a way that helps readers make sense of what’s happening and what the impact or reaction has been. On these platforms—and now Microsoft Start—readers can keep up with topics they actually care about instead of filtering out news that isn’t interesting to them. “I can see people who follow specific industries or specific niches or hobbies…I can see [AI-generated news feeds] being really useful and helpful,” Schmelzer added.
Generating News
Another way we could see news changing as a result of AI is actual AI-generated content, meaning technology has a hand in how content is created. “The idea of using AI to generate text is a field called natural language generation (NLG), which is part of the general market of NLP natural language processing. And I know that we’ve been tracking the use of NLG for news since, at the very least, 2016,” Schmelzer said. This doesn’t necessarily mean AI would replace professional journalists, but rather, AI would do the data-specific articles that don’t require a journalist’s skill set or creative writing. We’ve already seen this happening. In the 2016 Summer Olympics in Rio de Janeiro, an AI writing robot wrote about 58 articles per day about the games’ results. Even reputable new organizations like Forbes use a robot named Bertie to write data-specific stories about quarterly earnings reports. “I would say short-form, predictable format content will probably end up being algorithmically generated,” Schmelzer said. “That’s, I think, where the strength of algorithmically generated content is.”
