Archive: June 2020

Large language models (LLMs) pretrained on massive data are being used in countless real-world applications. However — as computer scientists have known for decades — not all data is equal, and this is also true with regard to the composition of LLM pretraining data, which is typically sourced from publicly available domains such as Wikipedia, book and web texts, etc. Most current data selection strategies for LLMs determine the mixture of domains (i.e. th

Illustration by Tag Hartman-Simkins / Futurism. Source: Getty Images A man who has been arrested on suspicion of starting the deadly Palisades Fire that destroyed thousands of homes and killed a dozen people in Los Angeles last year allegedly used ChatGPT to generate images of burning forests and cities, according to law enforcement officials. The suspect in custody, 29-year-old Jonathan Rinderknecht, was arrested near his home in Florida on Tuesday for de

Credit: Pixabay/CC0 Public Domain A new journal article by a researcher at the University of Manchester offers insight into how artificial intelligence (AI) is portrayed in leading newspapers worldwide, revealing a more nuanced and critical approach than previously assumed. New research, co-authored by academics from the University of Manchester and the University of Groningen, adds nuance to the idea that journalists have hyped AI technologies. Examining

Content warning: this story includes discussion of self-harm and suicide. If you are in crisis, please call, text or chat with the Suicide and Crisis Lifeline at 988, or contact the Crisis Text Line by texting TALK to 741741. Earlier this year, when OpenAI released GPT-5, it made a strident announcement: that it was shutting down all previous models. There was immense backlash, because users had become emotionally attached to the more “sycophantic” and war

Credit: Pixabay/CC0 Public Domain The number of police-involved lethal force incidents in the U.S. dropped 24% from 2021 to 2023, according to research from the Cline Center for Advanced Social Research at the University of Illinois Urbana-Champaign. The Cline Center's SPOTLITE project has compiled nearly a decade's worth of data to track and identify police uses of lethal force across the U.S. The SPOTLITE dataset is the most comprehensive nationwide acco

AI Generated Video (Screenshots). Source: Instagram; TikTok; YouTube Earlier this year, Netflix CEO Ted Sarandos bragged that generative AI is going to open the way for a paradigm shift in the way humans create art. “We’re confident that AI is going to help us and help our creative partners tell stories better, faster and in new ways — we’re all in on that,” he told CNBC in an interview. That might sound well and good in the board room. But here in the rea

Disney / Futurism Disney is planning to flood its streaming service, Disney+, with user-generated AI slop. During the company’s recent earnings call, Disney CEO Bob Iger said that the streaming service is “in the midst of rolling out the biggest and the most significant changes” since its inception six years ago, as quoted by The Hollywood Reporter. Specifically, Iger was referring to what sounds a bit like OpenAI’s Sora: a service that allows users to gen

Comparison of the strengths of humans and statistical machines, illustrating their complementary generalization in human–AI teaming scenarios. Credit: Nature Machine Intelligence (2025). DOI: 10.1038/s42256-025-01109-4 How do humans manage to adapt to completely new situations and why do machines so often struggle with this? This central question is explored by researchers from cognitive science and artificial intelligence (AI) in a joint article published

Getty / Futurism The AI industry continues to pour tens of billions of dollars into resource-intensive models and the infrastructure to run them. In the face of it all, their promises of kickstarting a technological revolution that could one day be immensely profitable remain convincing enough for investors to prop up sky-high valuations — at least for now. But actually turning a profit is likely many years away for the vast majority of AI companies, and t

Getty / Futurism We have it on good authority that next year is not 2027. That authority, of course, is Google’s horribly inaccurate AI Overview feature. By now, you’re probably well aware that the search engine’s AI-generated summaries are plagued with hallucinations. But you’d think after years of being unleashed on the unsuspecting public, they might be a little better. Fat chance, it turns out. Ask Google’s incredibly trustworthy AI model “is 2027 next

Credit: Materials & Design (2025). DOI: 10.1016/j.matdes.2025.114598 Imagine a fleet of submarines sitting idle on a military base in the Pacific because they contain malfunctioning or aging parts. Each submarine needs a unique replacement, available only from a single machinist more than 5,000 miles away. After months of waiting, the parts finally arrive—only for mechanics to discover a defect, and the process has to start over. This has been the reality

Published on March 4, 2026 at 11:03 Comment 1 𝕏 Print © Mariia Shalabaieva By opening the door of ChatGPT to the Pentagon, OpenAI has provoked the anger of its users and employees. Backed into a corner, the startup indicates it has made key changes to its agreement with the US military. OpenAI acknowledges, however, that its oversight on the use of AI in military operations is more than limited… Taking advantage of the discord between Anthropic and Trump's