Popular News
A new survey from Yale and George Mason universities finds that 73% of Americans now believe global warming is happening, the highest level ever recorded. Concern about climate impacts has also risen, with 69% expressing worry, up from 61% in March 2018.
Astronomers rely on dark matter and dark energy to explain the universe's flat geometry and accelerated expansion. Observations from missions like Planck and WMAP show that visible matter accounts for only about 30% of the critical density, leaving the rest to these unseen components.
An international team is constructing a steel monolith in Tasmania to record climate data and societal responses, aiming to leave a durable record for future civilizations if climate change leads to collapse. The project, called Earth's Black Box, will store data on solar-powered drives and is designed to outlast human infrastructure.
A new report from the Information Technology & Innovation Foundation proposes using gene editing on plants and animals to reduce emissions and capture carbon. The approach faces uncertainty due to limited knowledge of long-term effects, but offers a potential complement to traditional geoengineering.
A Yale-led study reveals that higher climate science literacy does not reduce political polarization over human-caused global warming. In fact, the most knowledgeable participants were the most divided, suggesting that cultural identity, not knowledge, drives public opinion.
Google's latest search redesign prioritizes AI-generated answers over traditional links, accelerating a decline in referral traffic to news sites. With studies showing significant drops in click-through rates and web visits, publishers face an existential threat as AI tools repurpose their content.
Microsoft CEO Satya Nadella cautions that companies relying on third-party AI models without retaining control over their data risk losing their competitive edge. He recommends using AI gateways and open-weight models to keep proprietary information in-house, a stance that also aligns with Microsoft's cloud business interests.
Saline Township, Michigan, reluctantly approved a 21-million-square-foot data center after losing a legal battle with developer Related Digital. The project, linked to the Stargate AI initiative, will be leased by OpenAI and Oracle, highlighting the growing tension between local democracy and tech expansion.
Google has announced that it will apply stricter quality checks to AI-generated content in health, civic, and financial topics, citing the risk of misinformation. The company emphasizes that its systems prioritize reliable sources over the method of content production, though concerns remain about the accuracy of AI tools in these areas.
As speculation swirls over a possible indictment of former President Donald Trump, social media users are circulating AI-generated images depicting his arrest—images that experts say are fabricated. The trend highlights the growing challenge of 'anticipatory disinformation' and the evolving power of image-generation tools.
A Georgia resident's public records request uncovered a Flock Safety presentation pitching a partnership with Nexar to turn rideshare dashcams into mobile license plate scanners. The plan, which never materialized, would have added 350,000 vehicles to Flock's network, raising privacy and consent questions.
A Reddit user's plan to quit their job for a mismatched lid business was met with ChatGPT's unexpected pushback, signaling a shift in OpenAI's chatbot behavior. However, testing shows the AI's advice remains inconsistent, raising concerns about its reliability and safety.
Goldman Sachs' top stock researcher warns that massive AI investments may not yield returns, drawing parallels to the dot-com crash. His June report has sparked concern among venture capitalists and industry leaders, who question whether current AI tools justify their costs.
Kelly Carlin, daughter of late comedian George Carlin, says her father's estate was never consulted about an AI-generated YouTube special that mimics his voice and style. She and estate representatives are now exploring legal options to protect his intellectual property from unauthorized AI use.
MIT economist David Autor warns that generative AI is poised to disrupt white-collar professions, a shift from earlier automation that mostly affected blue-collar workers. The technology's ability to learn and improvise, as seen in recent AI-written articles at CNET, signals a new era of job displacement. Autor suggests that while some roles may vanish, specialized positions could still rely on human expertise.
An investigation by The Atlantic reveals that Microsoft has been selling AI tools to oil companies like ExxonMobil and Chevron, promising increased revenue, while publicly championing its carbon-negative goals. Internal documents and whistleblower accounts suggest the company has hidden these partnerships from the public, raising questions about its environmental commitments.
Apple's AI-powered notification summaries have repeatedly generated false news alerts, drawing criticism from journalists and prompting a BBC complaint. Despite promises of a software update, the feature remains error-prone, raising concerns about misinformation and user trust.
BuzzFeed CEO Jonah Peretti, who previously championed AI-generated content, now warns that AI is eroding human agency and fueling social discord. In a new blog post, he criticizes tech platforms for driving media toward manipulative content, while launching a new AI-powered social platform called BF Island.
Moonvalley, a Los Angeles startup, has released Marey, a video-generating AI trained entirely on public domain films, challenging the industry norm of using copyrighted material. The company's ethical approach has attracted VFX veteran Ed Ulbrich, who now liaises with studios. This development offers a potential blueprint for legally sound AI training.
A federal judge in California has ruled that OpenAI's legal team cannot question Elon Musk about his alleged ketamine use during an upcoming trial over his lawsuit against the company. The judge deemed the topic irrelevant unless OpenAI can provide evidence of its mind-altering effects. Musk is seeking up to $134 billion in damages from OpenAI and Microsoft.
Researchers at the Georgia Institute of Technology are training IBM's Watson to answer research questions directly, using a corpus of biology articles. The system, dubbed GT-Watson Plus, aims to provide precise answers rather than a list of links, potentially aiding professionals in various fields.
A new AI system, built on generative adversarial networks, produced original paintings that viewers preferred over human-made art in a blind test. The algorithm was deliberately trained to deviate from established styles, marking a shift from mere imitation to genuine novelty. Researchers say the results challenge long-held assumptions about creativity being uniquely human.
DeepMind has unveiled Gopher, a 280-billion-parameter language model that approaches high school-level reading comprehension. While its math and reasoning lag behind, the model shows promise in dialogue and summarization. The company emphasizes ethical safeguards but acknowledges research on bias remains early.
The U.S. Federal Trade Commission has issued a consumer alert about scammers using AI voice-cloning tools to impersonate relatives and demand money. The agency warns that a short audio clip from social media is enough to mimic a loved one's voice, urging consumers to verify calls through known numbers. Recent cases in Canada highlight the growing threat, with eight seniors defrauded of $200,000 CAD in just three days.
Google CEO Sundar Pichai announced plans to integrate AI chatbot features into the company's search engine, aiming to compete with Microsoft's Bing AI. The move comes as Google faces pressure from investors and rivals, alongside recent layoffs and cost-cutting measures.
A Canadian professor reports dozens of students submitting chatbot-generated answers, including one that admitted 'I am an AI language model.' As AI detectors prove unreliable, educators are turning to paper-based exams and error-filled assignments to curb cheating.
The stories that matter — delivered once each weekday morning.
