Weekly Roundup: AI News Highlights
xAI’s Grok-1.5 Upgrade
Elon Musk’s AI venture, xAI, has unveiled the upgraded Grok-1.5 language model. The latest iteration boasts enhanced short-term memory and improved reasoning abilities. With a significant increase in short-term memory capacity, Grok-1.5 can now process up to 128,000 tokens, enabling better decision-making and insightful responses.
Language and Cultural Bias in AI
A study comparing LLMs developed in different countries on Traditional Chinese Medicine highlights the need for localized models. Chinese-developed LLMs outperformed Western ones in accuracy, emphasizing the importance of training models on local data.
DeepMind’s SAFE Fact-Checking App
DeepMind has developed SAFE, an AI-based app to fact-check LLM results. The system matches human fact-checkers’ accuracy and is now available on GitHub for public use.
LLM Applications in Azure AI
Microsoft’s Prompt Shields in Azure AI enhance content safety by protecting LLMs against prompt injection attacks. The system detects and prevents jailbreak attacks, ensuring the security and integrity of generative AI applications.
Adwise Partners Launches GambleAI
Adwise Partners introduces GambleAI, an LLM chatbot for iGaming. The innovative chatbot offers personalized and interactive user experiences, powered by leading LLM models.
Stay tuned for more AI updates and insights next week!
