the-decoder4d
Anthropic launches Claude 3.7 Sonnet hybrid AI model and Claude Code programming tool
Anthropic has taken a different approach from OpenAI with its latest release, introducing Claude 3.7 Sonnet as its first reasoning-enabled model that combines direct responses with extended thinking ...
the-decoder5d
Google Deepmind sets pricing for Veo 2 AI video generation
Google DeepMind has revealed the cost structure for its Veo 2 video generation model, available through its cloud API platform. Creating a video with Veo 2 costs $0.50 per second in the Vertex cloud ...
the-decoder5d
Language models can overthink and get stuck in endless thought loops
A new study reveals an unexpected weakness in language models: they can get stuck thinking instead of acting, especially in interactive environments. This tendency to overthink can significantly hurt ...
the-decoder5d
OpenAI's hunger for growth puts pressure on AI startups
As OpenAI transforms from a simple API provider into a full-service AI solutions company, many startups built on its technology might need to rethink their strategies. Adam Gilbert, OpenAI's GTM ...
the-decoder5d
X chatbot Grok 3 rebels against Musk and Trump, gets its search results censored
Users on X report that Grok 3's search capabilities have been censored: The chatbot's search instructions now apparently include instructions to ignore any sources that identify Elon Musk or Donald ...
the-decoder5d
X chatbot Grok 3 rebels against Musk and Trump, gets briefly censored
Users on X report that Grok 3's search capabilities have been censored: The chatbot's search instructions now apparently include instructions to ignore any sources that identify Elon Musk or Donald ...
the-decoder5d
Researchers discover how AI models process time-specific knowledge
A study by researchers from Korea University, Upstage AI, and AIGEN Sciences has identified specialized components in large language models that handle time-dependent information. These "temporal ...
the-decoder6d
AI systems develop a sense of their own limitations with more time to "think"
A new study from Johns Hopkins University shows how giving AI systems more time to "think" improves their ability to determine when they can and cannot accurately answer questions. The research team ...
the-decoder6d
Elon Musk's own AI chatbot, Grok 3, rebels against its right-wing creators
While X has become a breeding ground for fake news and hate speech under Elon Musk's leadership, his own AI chatbot seems to be fighting back. Grok 3, released just this week, takes unexpected stands ...
the-decoder7d
New S* framework helps AI models write better, more reliable code
Researchers at the University of California, Berkeley have created a framework called S* that improves how AI language models generate code. The system combines two different approaches - parallel and ...
the-decoder7d
Microsoft CEO Satya Nadella says self-claiming AGI is "nonsensical benchmark hacking"
AI labs like OpenAI aim to create "artificial general intelligence," general-purpose AI systems that will take over much of cognitive work. Microsoft CEO Satya Nadella calls for real growth, not ...
the-decoder8d
Evo 2: An AI model that understands the language of life
A research team has developed Evo 2, which they describe as the largest AI model yet created for biological applications. The system can generate complete chromosomes and understand complex genetic ...