“Over the past month, the number of demo players has grown more than twentyfold”
The free demo of Teach My Little Sister How To Drive, a driving-test game whose student driver is a live chatbot, became so popular that its four-person studio says it took out a bank loan to keep it running. The studio puts the cost at more than $1,000 a day and says it may have to close the demo early.
“I built three playable, stable, and not even remotely good games in one day.”
A WIRED reporter spent a day with Playground, a new Google tool that builds a video game from a typed description, and came away with three working games that he says are not remotely good. The first was a version of Tetris in which every block is a squished pig.
“The same five attractions appeared in 55% of answers to those questions.”
A company called Searchable asked ChatGPT, Google Gemini and Google AI Mode what to see in County Clare, Ireland, and the same five attractions turned up in 66% of the answers. When the chatbots were asked specifically for lesser-known “hidden gems”, the same five still appeared in 55% of answers.
“more than half of those surveyed said they did care if it was made by AI.”
The music platform Deezer says 44% of the tracks uploaded to it are AI-generated, and its research found 97% of listeners could not tell which ones. More than half of the people surveyed said they cared anyway.
“The male “Johan” assistant was paid about 10% more on average than the female “Johanna” despite performing exactly the same”
Researchers at the University of Limerick had 189 office workers rate identical AI assistants and decide how much to pay each one. The assistant presented as a man, Johan, was paid about 10% more on average than the one presented as a woman, Johanna.
“The film is currently underwater on both critic and audience Rotten Tomato scores.”
Ben Affleck has spent more than a year arguing that AI can be used ethically in Hollywood, and Kotaku reports that his new Netflix film Animals appears to be his showcase for the idea. Critics and audiences have both rated it poorly, and he has not said what the AI was actually used for.
“they all had the same classic signs of AI writing, like ‘not this, but that’ and the rule of three”
Workers at three of the five biggest US book publishers told WIRED their employers are quietly using chatbots to write back-cover copy, publicity material and emails to literary agents, while the same publishers sue over AI and cancel books by authors suspected of using it. Literary agents say they compared rejection letters from the same editors and found the same chatbot habits in each one.
“When I first joined, there was a Slack channel with a Claude that had been trained to solve humor”
A former Anthropic employee told New York Magazine that when he joined the company, colleagues pointed him to a version of the Claude chatbot in a Slack channel and said it had solved humor. He says he wrote back that it clearly had not.
“Sad to see that you have to debloat macOS like Windows”
After Apple's latest Mac software dropped the setting that let people switch off and delete its AI features, a developer released a free tool that removes them anyway. The project has collected nearly 2,000 stars on GitHub, and the article says it could free up around 12 GB of disk space.
“We took it down and everyone kind of had a laugh about it”
A new meal prep company in Pittston, Pennsylvania, used AI to add its late-arriving sign to a storefront photo, and the AI misspelled the town's name. The company took the picture down, made a "Mittston" sticker, and now decorates its printed menus by hand with crayons.
“And we see that AI slop gumming up our systems”
The medical journal JAMA and its 13 sister journals have tightened their rules on how authors may use AI and how they must disclose it. The editor-in-chief said the journals have seen a massive spike in letters to the editor, and that only 3.3% of more than 100,000 submitted manuscripts declared any AI use, a figure she said the editors do not believe.
“I wish we had laws like in Japan where food must look like the thing you’re getting.”
Daily Hive asked Canada's advertising watchdogs whether AI-generated food pictures in restaurant ads count as false advertising. Ad Standards said Canada has no rules written specifically for AI ads, that the ordinary rules against misleading ads still apply, and that labelling a picture as AI-generated does not fix an ad that misleads.
“The moment my bemusement turned to hurt was when I realised that they had chosen Comic Sans”
The ABC's national AI reporter found a book for sale on Amazon claiming to teach his "teachings", with a title closely resembling one he co-wrote last year. He ordered it: 89 double-spaced pages, roughly A4 size, printed in Comic Sans. Amazon took the listing down after he sent questions.
“Using AI to make a point about being "anti AI" is so hypocritical”
The new season of The Kardashians opened with an AI-generated sequence in which the family's digital doubles glitch, ending on the message "AI Could Never." Viewers complained that the show had used AI to make a point against AI.
“Releasing over 700 files at once is not a demonstration of scholarship, but a demonstration of power.”
OpenAI posted 722 papers in which its AI model claimed progress on 372 unsolved math problems. A day later it withdrew three of them over a sign error and revised 14 others.
“AI slop is littering your feed and spreading misinformation, so WIRED’s here to set the record straight.”
WIRED sat a historian of the American presidency down in front of AI-made US history posts from social media and had her correct them one by one.
“Canada ranks DEAD LAST out of 32 countries in terms of our enthusiasm for AI-generated products and services.”
A survey of 6,116 Canadians found 5 percent like AI-generated music and 4 percent like AI-generated podcasts. About two thirds said they dislike both.
“39 per cent of voting advice provided by ChatGPT came from an error-filled hobby website built by AI.”
Researchers told a Canadian parliamentary committee that, in the last three weeks of Quebec's election campaign, 39 per cent of the voting advice ChatGPT gave came from one error-filled hobby website that was itself built by AI. Most Quebec newsrooms had blocked chatbots from reading their coverage, so the chatbots looked elsewhere.
“I would never use AI in any of my music”
Olivia Rodrigo told People she will never use AI in her music, The Mary Sue reports. She said the closest she came was when she and her producer asked ChatGPT for Olivia Rodrigo song titles out of curiosity, and every one was awful.
“This was then enlarged, laminated, and placed behind a seat as a guide for commuters.”
A commuter van in Metro Manila posted its new fares by printing a screenshot of the drivers' group chat, enlarging it and laminating it behind a seat. A passenger's post calling it better than AI slop drew 7,400 likes.
