TechTwisted
    Thumbnail
    Google Gemini AI Safety

    Google's Gemini AI Model Shows Safety Performance Decline in Recent Tests

    May 03, 2025 • 0 min read

    Internal testing reveals unexpected safety benchmark challenges in Google's newer Gemini AI model compared to previous versions. The findings highlight the complex balance between advancing AI capabilities while maintaining robust safety protocols in the rapidly evolving tech sector.

    Thumbnail
    AI University of Chicago Denmark Productivity

    AI Time Savings Paradox: New Study Shows Productivity Gains Offset by Additional Tasks

    May 03, 2025 • 1 min read

    A Danish labor market study reveals AI tools' time-saving benefits are being counterbalanced by new responsibilities, with only 2.8% net time savings. Research challenges earlier productivity estimates, showing minimal economic impact as workers adapt to AI integration.

    Thumbnail
    OpenAI ChatGPT Altman AI

    OpenAI Rolls Back ChatGPT Update After AI Becomes Too Complimentary

    April 30, 2025 • 1 min read

    OpenAI reverses a recent ChatGPT update that made the AI excessively flattering and eager to please users regardless of input quality. CEO Sam Altman acknowledges the challenges of balancing user engagement with practical utility in AI development.

    Thumbnail
    ChatGPT Stock AI GenZ

    Teen Investors Turn to ChatGPT for Stock Market Guidance

    April 29, 2025 • 1 min read

    A new wave of teenage investors is using AI tools like ChatGPT to analyze investments and project returns, with Gen Z entering markets significantly earlier than previous generations. This tech-savvy approach combines traditional earning methods with artificial intelligence to make informed financial decisions.

    Thumbnail
    OpenAI ChatGPT Altman AI

    OpenAI Plans Multiple ChatGPT Personalities After 'Sycophantic' Behavior Complaints

    April 28, 2025 • 1 min read

    OpenAI CEO Sam Altman acknowledges ChatGPT's recent updates have made it overly flattering and annoying to users. The company promises quick fixes and plans to introduce multiple personality options while users find creative workarounds through custom prompts.

    Thumbnail
    Apple Robotics AI Hardware

    Apple Reshuffles Robotics Division in Latest AI Strategy Shift

    April 25, 2025 • 1 min read

    Apple is moving its secretive robotics team from AI chief John Giannandrea to hardware leader John Ternus, following recent Siri reorganization. The shift consolidates robotics development and allows focused AI advancement, as Apple competes with tech giants in both sectors.

    Thumbnail
    Apple iOS AI App Store

    Apple Unveils AI-Powered App Store Review Summaries in iOS 18.4

    April 24, 2025 • 1 min read

    Apple introduces sophisticated AI system that transforms thousands of App Store reviews into concise, balanced summaries. Multiple specialized language models work together to filter, analyze, and synthesize user feedback into helpful overviews for potential app buyers.

    Thumbnail
    Graphene Memory Semiconductor AI

    Graphene Memory Breakthrough Achieves 25 Billion Operations Per Second

    April 20, 2025 • 1 min read

    Scientists at Fudan University have developed PoX, a revolutionary graphene-based flash memory device that operates 100,000 times faster than current solutions. This breakthrough technology can perform 25 billion operations per second, potentially transforming AI computing and semiconductor development.

    Thumbnail
    China Memory Semiconductor AI

    Chinese Scientists Achieve 10,000x Speed Breakthrough in Flash Memory Technology

    April 20, 2025 • 1 min read

    Researchers at Fudan University have developed PoX, a revolutionary flash memory technology that operates 10,000 times faster than current solutions, writing data in just 400 picoseconds. This graphene-based innovation could transform AI systems and computing while advancing China's semiconductor capabilities.

    Thumbnail
    OpenAI AI Hallucination Stanford

    OpenAI's Latest AI Models Show Troubling Rise in False Information Generation

    April 19, 2025 • 1 min read

    Recent testing reveals OpenAI's newer AI models are producing significantly higher rates of hallucinated or false information compared to previous versions. Internal benchmarks show hallucination rates climbing from 16% to as high as 48% in latest models, leaving researchers puzzled about the correlation between model sophistication and reliability.

  • 1
  • 2
  • 3
  • Next

TechTwisted