gekko

doing the math for you

Author: gekko

  • Teaching LLMs to Ask Smarter Questions: Bayesian Experimental Design for Multi-Turn Information Gathering

    Teaching LLMs to Ask Smarter Questions: Bayesian Experimental Design for Multi-Turn Information Gathering

    —

    by

    gekko
    in

    Large Language Models (LLMs) have shown remarkable capabilities in understanding and generating text, but they struggle with adaptive, multi-turn information gathering – i.e. asking relevant follow-up questions based on previous answers. The paper “BED-LLM: Intelligent Information Gathering with LLMs and Bayesian Experimental Design” (arXiv:2508.21184) addresses this shortcoming by introducing a new approach that enables LLMs…

  • AI Companions in the Classroom: Parallels Between Gibson’s Vision and the Educational AI Revolution

    AI Companions in the Classroom: Parallels Between Gibson’s Vision and the Educational AI Revolution

    —

    by

    gekko
    in

    In William Gibson’s 1988 novel Mona Lisa Overdrive, the third installment of his Sprawl trilogy, young Kumiko Yanaka receives a sleek, wand-like device from her Yakuza father before being sent to London for safety. This device houses an advanced AI named Colin—a sophisticated, invisible companion who serves as guide, confidant, and protector, manifesting only to…

  • Implications of AGI for Universal Basic Income

    Implications of AGI for Universal Basic Income

    —

    by

    gekko
    in

    The emergence of artificial general intelligence (AGI) – AI systems capable of matching or exceeding human abilities across virtually all tasks – promises to profoundly reshape economies and societies. As machines become able to perform any job a human can do, the traditional link between employment and livelihood comes into question. In a post-AGI world,…

  • AI Psychosis and Tech Obsession: New Delusions, Old Patterns

    AI Psychosis and Tech Obsession: New Delusions, Old Patterns

    —

    by

    gekko
    in

    In recent months, reports have emerged of AI chatbots seemingly driving some users into delusional or harmful mental states. The phenomenon, informally dubbed “AI psychosis,” describes instances where people develop hallucinations, paranoid ideas, or extreme emotional attachments as a result of intensive conversations with large language model (LLM) chatbots. This is not happening on a…

  • Less Is More: When Overthinking Leads to Worse Outcomes

    Less Is More: When Overthinking Leads to Worse Outcomes

    —

    by

    gekko
    in

    In both human behavior and AI systems, more effort can paradoxically produce worse results. (Anthropic’s recent Inverse Scaling in Test-Time Compute paper demonstrated that giving an AI model more “thinking” time can degrade its performance.) Likewise, psychology and behavioral research reveal many situations where “less is more” for humans—extra thought, choices, or analysis actually impair…