All Categories
Nebulatitan
BuzzFeed CEO Warns of AI's Threat to Human Agency After Years of AI Content
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.
Court Bars OpenAI from Questioning Musk on Drug Use in Trial
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.
Apple's AI Summaries Draw Fire for False Alerts on News Apps
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.
Carlin Estate Weighs Legal Action Over AI Voice Clone in YouTube Special
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.
White-Collar Jobs Face AI Disruption, MIT Expert Says
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.
DeepMind's Gopher Model Nears Human-Level Reading Skills in New Tests
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.
Viral AI Cover Reimagines Taylor Swift Hit in Johnny Cash's Voice
A Dallas musician's AI-generated cover of Taylor Swift's 'Blank Space' performed in the style of the late Johnny Cash has gone viral, sparking conversations about the ethics and creativity of voice cloning. The 53-second clip, posted on TikTok, has drawn widespread attention for its authentic sound despite Cash's death over two decades ago.
College Students Caught Copy-Pasting AI Chatbot Responses
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.
Google CEO Confirms AI Chatbot Integration in Search Engine
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.
AI Voice Cloning Scams Prompt Federal Alert as Fraudsters Target Families
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.
Machines Outpaint Humans in New Turing-Style Test for Creativity
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.
Georgia Tech Researchers Teach IBM Watson to Answer Research Questions
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.
Scion Asset Management Winds Down as Burry Exits Short Positions
Michael Burry, famed for his 2008 housing market short, is liquidating Scion Asset Management after a brief, high-profile bet against AI stocks. The move, revealed in an investor letter, has sparked speculation about his next steps and the state of the AI market.
AI Reimagines Times Square Footage as a Cubist Masterpiece
A new AI video transforms Manhattan footage into a cubist painting, showcasing the power of style transfer. The project, posted by YouTube channel Matchue, blends video content with Carlos Merida's 1982 artwork to create a surreal urban landscape.
Robots Join Real Estate: New Tools for Buying and Selling Homes
Real estate firms are deploying robots to handle tasks like client communication and property tours, potentially lowering costs. While some see this as a disruption, industry leaders debate whether robots will replace human agents.
Doctors and Patients Embrace AI Despite Thin Proof of Real-World Benefit
A Nature Medicine editorial warns that AI tools for medical advice are being adopted faster than evidence supports, citing scarce proof of patient benefit and persistent flaws like hallucinations. The journal calls for clear evaluation standards and benchmarks before clinical claims are accepted.
OpenAI's GPT-4 Technical Report Draws Criticism for Lack of Transparency
OpenAI's release of GPT-4 included a 98-page technical report that omits key details about the model's architecture and training. Critics, including Lightning AI CEO William Falcon, argue this lack of transparency sets a dangerous precedent for AI development and regulation.
Meta Removes Posts Offering Abortion Pills by Mail, Activists Say
Meta-owned Facebook has been deleting posts that offer to mail abortion pills, with some users receiving temporary bans. The removals occur even though mailing such pills is legal in most states, raising concerns about censorship and public health access after Roe v. Wade was overturned.
Meta's AI Brain Drain Accelerates as Three Top Researchers Depart in March
Meta has lost three senior AI researchers in March, including a director of generative AI and the head of FAIR. The departures come amid intense competition and a venture capital boom in AI, prompting experts to see it as part of a broader Silicon Valley cycle.
Sci-Fi Magazine Overwhelmed by AI-Generated Submissions
Clarkesworld, a leading sci-fi magazine, temporarily halted submissions in February 2023 after receiving a flood of AI-generated stories. Editor Neil Clarke built a spam filter with volunteer help, but warns that the current situation is unsustainable without industry-wide solutions.