Welcome back to AI Hungry, where the pulse of artificial intelligence innovation beats with every story. Dive into our latest update to discover how AI is reshaping the German workplace, with a significant number of office workers now relying on AI tools weekly. Then, shift your attention to Meta's groundbreaking release of Llama 3, a language model that's setting new standards in the open-source AI community.
Our coverage spans from practical applications affecting today's workforce to the frontiers of AI research and development. Whether it's the integration of AI in daily tasks or the unveiling of pioneering language models, these articles are poised to inform and intrigue our AI-savvy readership.
One in Five German Office Workers Utilize AI Weekly, Reveals Survey
A recent survey by Microsoft and Civey indicates that 21.5% of office workers in Germany use AI tools at least once a week, showcasing the growing integration of AI in the workplace. The survey, which included 9,000 respondents, also reveals that 28% of Germans believe AI boosts productivity and one in five see it aiding in time management. However, there is still uncertainty among employees about whether AI represents an opportunity or a risk for their jobs.
Despite the increasing use of AI, there is a notable gap in knowledge about the technology, with only 23.9% of office workers and 18.5% of industrial workers feeling well-informed. Over half of the general public admits to poor AI knowledge, yet interest is high, particularly among the younger demographic, who are receptive to AI's potential in various aspects of life.
Meta Launches Llama 3: A New Benchmark in Open-Source Language Models
Meta has introduced Llama 3, a cutting-edge open-source language model that surpasses its predecessors and competitors in various tasks such as reasoning, code generation, and instruction following. Llama 3 is acclaimed as the most advanced large language model available openly, outperforming similar-scale models from Google and Anthropic. It comes in two sizes, with the smaller model having eight billion parameters and the larger one boasting 70 billion parameters. Both models can process inputs of around 6,000 words and are available for download or on cloud services like AWS.
The development of Llama 3 is part of Meta's ongoing efforts to enhance AI capabilities, with plans to release even larger models in the future, including a 400 billion parameter version. The new models are designed with improved architecture for better text generation and efficiency, and they are equipped with tools for responsible deployment, such as content filtering and risk assessment features. Llama 3 is also powering Meta AI's new chatbot solution, currently available in English in select countries.
🤖 Meta AI Expands Reach Across Apps and Launches Llama 3 Model. Meta AI, now powered by the Llama 3 model, extends its integration into Meta's apps and launches a standalone website. The AI can assist with tasks like planning trips within Messenger and enhancing Facebook feed interactions. (Link)
🚀 Salesforce Enhances Slack with New Generative AI Features. Salesforce updates Slack with generative AI capabilities, offering users daily recaps and expanding language support. The move aims to democratize AI at the staff level and improve productivity. (Link)
🚀 Startup Groq's AI Chip Hits Remarkable Speeds with Meta's LLaMA 3 Model. Startup Groq's new AI chip architecture has achieved over 800 tokens per second with Meta's LLaMA 3 model, potentially outperforming current cloud AI services and challenging Nvidia's market dominance. (Link)
🤖 AI Chatbot Grok Spreads Fake News Based on User Jokes. AI chatbot Grok mistakenly turns user jokes into fake news, like NBA player Klay Thompson damaging homes with bricks and bizarre reports about solar eclipses, highlighting the risks of generative AI without proper safeguards. (Link)
🔒 Major Deepfake Porn Sites Block UK Access Ahead of New Law. Following the UK's announcement of a law criminalizing nonconsensual deepfakes, two major deepfake pornography websites have blocked UK users. This move signifies a significant step in combating deepfake abuse. (Link)
✈️ AI Pilot Successfully Engages in Dogfight with Human-Piloted F-16. An AI pilot developed by DARPA has successfully engaged in a dogfight with a human-piloted F-16, demonstrating significant advancements in autonomous aerospace technology. The event marks a breakthrough in AI applications for defense and raises important ethical considerations. (Link)
