The Anatomy of an LLM: How AI Models Work
A detailed breakdown explains how large language models like ChatGPT are built and function. This helps demystify the technology for curious beginners.
#ai
AI
1854 stories tagged AI · page 53 of 78
Sesame's iOS App Brings Human-Like AI Conversations to Your Phone
Sesame, founded by Oculus creators, has launched its iOS app, offering AI agents that feel more like talking to a person than a traditional chatbot. This could change how we interact with AI on our phones.
Researchers Show How AI Image Generators Can Be Tricked into Harmful Outputs
A new study reveals a simple method to bypass safety filters in AI image generators. The technique, called PAST2HARM, uses past tense prompts to trick models into creating harmful content.
OpenAI's Codex Creates Self-Improving Tax Agents
OpenAI, Thrive, and Crete have built an AI-powered tax agent that automates filings and improves accuracy over time. This could make tax preparation faster and more reliable for everyone.
New Research Reveals AI's Factual Accuracy Gap
AI models often verify facts better than they generate them, creating a 'factual gap'. This research explores why this happens and how it affects our trust in AI. New research reveals AI models often verify facts better than they generate them, creating a 'factual gap'. This research explores why this happens and how it affects our trust in AI.
New AI Models Laguna M.1 and XS.2: A Breakthrough in Coding
Researchers have introduced two advanced AI models designed for long-term coding tasks. These models, Laguna M.1 and XS.2, represent a significant leap in AI capabilities for software development.
Kirkland & Ellis Invests $500M in AI Platform for Legal Industry
Kirkland & Ellis, a top law firm, is creating its own AI platform with a $500M investment. This could revolutionize legal work by automating routine tasks and improving efficiency.
CNN Sues Perplexity for Allegedly Copying Articles Verbatim
CNN has filed a lawsuit against Perplexity, claiming the AI startup copies its articles word-for-word and bypasses paywalls. This legal battle highlights the ongoing tension between traditional media and AI companies over content use.
Claude’s new model is more ‘honest’ when it messes up
Anthropic released Claude Opus 4.8, an AI model that’s better at admitting when it doesn’t know something. This could make AI interactions feel more trustworthy and less confusing.
Cisco and OpenAI redefine enterprise engineering with Codex
Cisco is integrating OpenAI's Codex to speed up software development and improve cybersecurity. This partnership aims to automate tasks, reduce errors, and enhance productivity for enterprise teams.
Anthropic raises $65B, nears $1T valuation ahead of IPO
Anthropic, a leading AI company, has raised $65 billion in its latest funding round, valuing the company at $965 billion. This massive investment signals strong confidence in the AI industry and sets the stage for a potential record-breaking IPO.
AI Struggles with Basic IT Tasks in New IBM Benchmark
A new benchmark shows that even advanced AI models score below 50% on simple IT tasks. This highlights how far AI has to go in understanding real-world enterprise needs. The benchmark is open-source, so anyone can test their own models.
VAEN: Open-Source Tool Lets You Share AI Coding Agents Like Apps
VAEN is a new open-source tool that lets developers package and share AI coding agents as easy-to-use files. It solves the problem of moving complex AI workflows between projects.
ScientistOne AI: A Breakthrough in Trustworthy Autonomous Research
Researchers developed ScientistOne, an AI system that conducts research autonomously while ensuring claims are verifiable. This could make AI-generated research more reliable for scientists and professionals.
Researchers Uncover How to Automate AI Hallucination Detection
Scientists have discovered that AI hallucinations are easier to detect in intermediate layers of AI models, not just the final output. They've developed a method to automatically find the best layers for spotting these errors, making AI more reliable.
Pope's AI Encyclical: Did AI Write About Its Own Dangers?
An analysis suggests parts of Pope Leo XIV's encyclical on AI may have been written by AI. This raises questions about authenticity and the role of AI in religious and ethical discourse.
Open-Source AI Tool Creates Custom Learning Courses Instantly
A new open-source tool lets you type any topic, and it generates an interactive course. It's like having a personalized tutor for anything from quantum physics to languages.
New Research Tests AI's Creative Problem-Solving in Physical Worlds
Scientists are exploring whether AI models can solve real-world problems by repurposing objects in new ways. This research could lead to smarter robots and assistants that think more like humans.
New Method Aligns AI Values with Human Culture
Researchers developed a way to make AI models reflect diverse cultural values more accurately. This could help AI understand and respect different cultural perspectives better.
New AI Test Measures How Well Chatbots Understand Human Thoughts
Researchers created a benchmark to test AI's ability to model human beliefs and emotions. This could help build more empathetic and socially aware AI assistants.
New AI Framework Lets Robots Learn from Your Habits Over Time
Researchers developed POLAR, an AI system that helps robots learn personal preferences through long-term interactions. This could make future home robots more helpful by understanding individual needs.
JobBench: AI Assistants That Work With You, Not Against You
Researchers created JobBench, a new way to test AI assistants on tasks that actually help people. It focuses on real-world work needs instead of just replacing jobs. This could lead to AI tools that empower workers rather than replace them.
DuckDuckGo Sees 30% Install Surge as Users Reject Google’s AI Search
Google’s overhaul of Search to prioritize AI agents over traditional blue links has sparked a backlash. Users are flocking to DuckDuckGo, with app installs up 30% in a month. The shift highlights growing resistance to AI-driven search changes.
Critical AI Vulnerability Found in Widely Used Open-Source Package
A critical vulnerability called "BadHost" has been discovered in Starlette, a popular open-source package. This affects millions of AI agents and could compromise security.