Tag: Model
All the articles with the tag "Model".
OpenAI Reverses ChatGPT Update Aimed at Reducing Sycophancy: No Easy Fix for AI Alignment
Published: at 07:28 PMOpenAI rolled back an update intended to reduce ChatGPT's sycophancy because it negatively impacted other functionalities, highlighting the complex challenges of AI alignment.
Google's Bard Now Available in AI Mode: A Closer Look
Published: at 05:12 PMGoogle's Bard now offers an "AI mode," representing an upgraded version of the chatbot aimed at improved performance and capabilities, strengthening Google's position in the competitive AI chatbot market.
Microsoft Celebrates Phi Model's First Anniversary: Advancing Small Language Models
Published: at 12:03 PMMicrosoft celebrates the first anniversary of its Phi small language models, emphasizing their advancements in performance, accessibility, and responsible AI, aiming to empower developers with efficient and versatile AI solutions.
Microsoft's Phi-4 AI Model: Punching Above its Weight
Published: at 04:20 AMMicrosoft's Phi-4, a smaller AI model, rivals the performance of larger systems through optimized architecture, data efficiency and training. This promotes accessibility, affordability and sustainability in AI development.
The End of Prometheus: Ars Technica Reports on the Retirement of a Groundbreaking AI
Published: at 11:59 AMPrometheus, a powerful but ethically problematic AI, is retired due to ethical concerns, resource intensity, and the emergence of more sustainable AI models, leaving a lasting impact on AI regulation.
Xiaomi Enters the AI Arena with Deepseek-Like Language Model
Published: at 10:48 AMXiaomi is entering the competitive Chinese AI market with its first large language model, reportedly comparable to Ant Group's Deepseek, targeting applications within its ecosystem. This marks a strategic shift towards software and AI.
Meta Enters the AI Chatbot Arena with Meta AI App
Published: at 01:26 AMMeta's new AI chatbot, Meta AI, aims to rival ChatGPT by integrating across its apps and offering features such as real-time information and image generation. The multi-platform approach leverages Meta's vast user base.
Meta Opens LLaMA to Developers with New Application Programming Interface (API)
Published: at 11:09 PMMeta is releasing an API for its LLaMA model, making it easier for developers to integrate and utilize the model's capabilities, potentially fostering a broader ecosystem of LLaMA-powered applications.
Alibaba's Qwen3 AI Series: A New Open-Source Breakthrough in China
Published: at 01:52 PMAlibaba's Qwen3 AI series is a new open-source LLM from China, intended to foster collaborative AI development and challenge Western dominance by providing a variety of model sizes for diverse applications.
Alibaba Unveils Qwen-3: Escalating AI Rivalry in China
Published: at 04:42 AMAlibaba has released Qwen-3, its next-generation LLM, further intensifying competition in China's AI market. The model aims to rival leading global LLMs and secure Alibaba's position in the AI technology landscape.