In-depth articles with more detailed investigations, interesting thought pieces, and more.
China’s ByteDance is transforming data from its popular apps TikTok, Douyin, and Toutiao into a competitive advantage in artificial intelligence, reports Meaghan Tobin of The New York Times. The company collects behavioral data from approximately 170 million U.S. TikTok users and around 1 billion users of its Chinese apps. This wealth of information has become …
Google is leading the AI race with its Gemini 2.5 Pro Experimental model, which currently ranks as the best performing AI model across multiple benchmarks. According to Alberto Romero in his newsletter The Algorithmic Bridge, Google now dominates on every AI front. The model tops leaderboards including LMArena, GPQA Diamond, and Humanity’s Last Exam, outperforming …
OpenAI has significantly shortened its safety testing period for new AI models, prompting concerns about insufficient safeguards. According to a Financial Times report by Cristina Criddle, testers now have just days to evaluate models compared to several months previously. Eight people familiar with OpenAI’s testing processes indicated that evaluations have become less thorough as the …
A new report from The Information reveals significant internal turmoil behind Apple’s attempts to revamp Siri. According to a post by Hartley Charlton for Macrumors, Apple’s AI team repeatedly changed direction on the technical approach for Apple Intelligence, starting with plans for dual language models before pivoting to a single cloud-based solution. The report highlights …
James Cameron, director of films like “Avatar” and “Titanic,” sees potential in AI technology to reduce filmmaking costs significantly. In a podcast interview with Meta CTO Andrew Bosworth, Cameron expressed that generative AI could cut the cost of blockbuster films by half, primarily by increasing production efficiency rather than reducing staff. As Alex Weprin reported …
Researchers from SentinelOne’s SentinelLabs have discovered that spammers utilized OpenAI’s chatbot to generate unique messages that successfully bypassed spam filters on more than 80,000 websites over a four-month period. According to a report published by Dan Goodin on Ars Technica, the operation went undetected by OpenAI for at least four months before the company revoked …
Website owners are experiencing dramatic traffic losses following Google’s introduction of AI Overviews, according to an investigation by Davey Alba and Julia Love of Bloomberg. The report, based on interviews with 25 publishers, reveals that many independent websites have seen traffic plummet by up to 70% since March 2024, disrupting the long-standing relationship between content …
Stanford University’s 2025 AI Index reveals that AI training costs have reached unprecedented levels, with Google’s Gemini 1.0 Ultra costing approximately $192 million to train. The comprehensive report, analyzed by Eliza Strickland for IEEE Spectrum, highlights both challenges and progress in the AI field through 12 significant graphs. Despite rising training expenses, the cost of …
A new Pew Research Center report reveals significant gaps between how the American public and AI experts perceive artificial intelligence. According to researcher Colleen McClain’s findings, while experts are largely optimistic about AI’s impact (56% positive), only 17% of the general public shares this enthusiasm. The surveys, conducted in 2024, show that AI specialists are …
Midjourney, primarily known for AI image generation, has released new research in collaboration with New York University on training large language models to produce more creative text. Carl Franzen reports for VentureBeat that the research introduces two new techniques: Diversified Direct Preference Optimization (DDPO) and Diversified Odds Ratio Preference Optimization (DORPO). These methods encourage LLMs …