People Inc bets human-created recipes can beat LLMs | OpenAI must share 20m chat logs with publishersAnd News Not Slop campaign launches as Politico journalists celebrate victory against AI in their newsroomWelcome to this week’s Press Gazette Future of Media US newsletter on Friday, 5 December. 🥘I just searched “vegan pie recipe” on Google and was served an AI Overview giving me the full ingredients and method for a mushroom and potato pie. At first glance, without cooking it, the recipe appears to make sense and not have any Rachel-from-Friends meat trifle-level errors. That appears to be because the recipe is exactly copied - albeit with some crucial information like cooking times taken out - from a BBC recipe. But why would the average user bother to click through to the BBC when a full recipe is right there with no extra effort needed? This has been my concern for a while regarding publishers of recipe content. It feels like it’s more at risk from AI summaries than news articles in particular. But People Inc appears to disagree. It launched a new brand MyRecipes this year and has gone from zero to two million registered users in six months. “We are called People Inc, after all, and we think there’s something to recipe content that is very much a human experience and can’t necessarily be replicated with AI.” 👩⚖️ This week a judge (again) ordered OpenAI to hand over 20 million anonymised ChatGPT user conversations to news publishers in their ongoing legal fight. OpenAI has been up in arms about the idea of doing so, calling it an “overreach on user privacy” despite the massive de-identification process going on. But, as the judge questioned, why was it so far through the anonymisation process if it was never planning to hand over the information? I loved this quote from Frank Pine, executive editor of MediaNews Group and Tribune Publishing: “OpenAI’s leadership was hallucinating when they thought they could get away with withholding evidence about how their business model relies on stealing from hardworking journalists…” 🤖 And we looked at a huge moment for Politico journalists and AI in the newsroom more widely as an arbitrator said the use of AI-generated live summaries on the homepage without human editing, containing errors, breached the Axel Springer brand’s union contract. The Politico victory came as the News Guild launched the News Not Slop campaign. The journalists’ concerns over the damage AI can do to the credibility of their newsrooms is a completely legitimate - and urgent - one. In an industry that has already seen trust slipping for years now, why would we hasten it along? Once trust is lost, even through the publication of simple and seemingly unharmful errors, it’s a much longer slog to try and get it back again. 📈On Press Gazette this week:1) Dubious experts deployed by MyJobQuote published more than 600 times in UK pressQuoted experts do not appear to be real and some have AI-generated profile pictures. 2) Future takes action on ‘Google Zero’ as revenue declinesFuture focusing on engaging audience without relying on Google. |