Most of us by now have become more or less desensitized to a steady drip of AI slop: AI-generated influencers, musicians racking up millions of streams, LinkedIn posts—a sizable chunk of online content is now being churned out by bots, without any kind of federal regulation to keep it in check.
Even the news is becoming automated: As Wired reported Wednesday, an AI-driven publication called RuntimeWire was the first to publish a major scoop about the recent hack into Hugging Face by rogue OpenAI models, beating actual tech publications to the punch by a significant margin.
That obviously raises big questions around the future of journalism at a time when the industry is already struggling to contend with the rise of AI chatbots and their siphoning of traffic from traditional online search, which has long served as the backbone of digital publishing.
The theory behind RuntimeWire
Much of the job of human reporters is pattern-recognition: spotting trends and connecting dots quicker than your competitors. That means that LLMs, with their superhuman ability to detect signals from oceans of noisy data, are at least theoretically well-positioned to spot story ideas that might elude human journalists. (An algorithm can scrutinize a companies’ tax records much more meticulously and quickly than a human can, for example.)
That’s the idea behind RuntimeWire, “an AI-native startup news and intelligence publication built for founders, operators, investors, and engineers,” according to its LinkedIn page. Launched in May by Austin-based entrepreneur and tech investor Ryan Merket, RuntimeWire is positioned as a new kind of newsroom that blends the speed of AI with at least some degree of human editorial oversight. Think of it like informational fast food: speediness at the expense of quality.
“We are a lean team covering a beat that used to take a much larger one, so we lean hard on modern tools—including AI—to move faster,” the company wrote in the “Ethics” section of its website, adding that it “uses AI to monitor sources, surface what is genuinely new across thousands of feeds, transcribe, summarize long documents, and handle the grunt work of research… But the pipeline starts and ends with people.” The site also includes AI-generated header images for stories, as well as a “Daily Recap” podcast with automated hosts. RuntimeWire is currently hiring for a contract-based “Startup News Writer” position, according to its website. One of the listed requirements says that applicants must be “comfortable being assisted by AI tools.”
RuntimeWire’s AI-powered approach meant that it was able to publish an article about new revelations made by OpenAI employees during a recent conference about the notorious Hugging Face hack quicker than human journalists, including Wired’s Lily Hay Newman. As you could probably guess, the tone of the AI-generated article (which like most of the stories published on the site includes Merket’s name in the byline) is rather flat and mechanical compared with those penned by human authors. It reads like a play-by-play account of what happened, with none of the between-the-lines insights or investigative depth that can be expected from many human-run newsrooms.
Merket told Gizmodo that while scrolling through X, he spotted a photo shared by OpenAI president Greg Brockman containing a QR code to the conference’s live caption stream. He captured a transcript of the stream and “then submitted that original source material into RuntimeWire’s reporting system, which researched, drafted, edited, and fact-checked the story.”
Looking ahead
Since RuntimeWire is only a few months old, it’s too soon to say with any certainty what its prospects are for long-term success. (The company’s articles were read over 118,000 times last month, according to an announcement published to its site.)
What does seem certain, however, is that AI-driven newsrooms will proliferate as models themselves get better at writing, aggregating information, and minimizing hallucination. And given how quickly the culture at large seems to have become inured to the growing presence of AI-generated content across the internet, it’s easy to imagine a not-so-distant future in which a portion of the population would rather get cut-and-dry news from bots quickly, rather than wait a little longer for more in-depth reporting written by humans. Others will probably prefer to pay for news from outlets that make it a point to not include AI in any point of the editorial process. For many people, the speed of informational fast food will be too good to pass up. For the rest of us, it will only make home-cooked meals taste that much better.
Read the full article here
