Meta Unveils Llama 3: A New Era of Open Source AI and Stand-Alone Chatbot
Meta has launched Llama 3, the latest iteration of its large language model, alongside a new stand-alone Meta AI chatbot. Llama 3, which is now among the most powerful open-source AI models, comes in two versions: an 8 billion and a 70 billion-parameter model, with a 400 billion parameter model currently in training. Ragavan Srinivasan, Meta’s VP of Product, highlighted the exceptional performance of the upcoming 400 billion parameter model. The initial releases, the 8B and 70B versions, have shown remarkable benchmarking capabilities, outperforming rival models from Google, Anthropic, and Mistral in various aspects. Meta has also introduced a new AI chatbot, positioning itself as a direct competitor to ChatGPT, Anthropic’s Claude 3, and HuggingFace’s HuggingChat. This chatbot, as stated by Meta CEO Mark Zuckerberg, aims to be the most intelligent AI assistant available for free use. It features Meta Imagine, an image generation model, and can surface live information from the web, including Microsoft’s Bing and Google search results. Despite its advancements, Llama 3 is not yet multi-modal, but plans for such a version are underway. The model is available under a Meta-specific license, sparking discussions within the open-source community about its true openness.
Salesforce Enhances Slack with AI-Driven Recaps and Multilingual Support
Salesforce continues to innovate, rolling out significant updates to Slack that enhance its generative AI capabilities, known as Slack AI. These updates include smart recaps and expanded language support, marking a significant leap in making generative AI accessible at the staff level. Introduced initially in February, Slack AI features aimed at simplifying work interactions through AI search and summaries have now been augmented with a Recap feature. This allows users to receive daily digests of discussions from selected channels, leveraging AI to generate summaries and relevant citations. The Recap feature is designed to help users stay updated without the need to navigate through each channel individually. Salesforce is also broadening the linguistic accessibility of Slack AI, adding Japanese and Spanish to its repertoire, with plans to include more languages in the future. The update also hints at future enhancements, including the integration of more data sources like files, apps, canvases, and clips, to further refine AI capabilities within Slack. Salesforce is exploring the use of large language models for analyzing audio and video interactions, aiming to provide AI-generated summaries of key discussions and action items.
Meta Integrates AI Chatbot Across Platforms Amid Misinformation Concerns
Another update from Meta. Meta is set to integrate its conversational chatbot, Meta AI, across its array of social media platforms, including Facebook, WhatsApp, Instagram, and Messenger. Announced by CEO Mark Zuckerberg, this move aims to enhance user interaction by enabling the chatbot to generate images and answer questions. This development coincides with the introduction of Meta’s latest large language model, Llama 3, positioning Meta AI in direct competition with leading AI chatbots like OpenAI’s ChatGPT and Google’s Gemini. However, this advancement has sparked concerns among experts about the potential for increased misinformation, hate speech, and extremist content on Meta’s platforms. The company’s image generator, in particular, has raised questions regarding its depiction of race and gender. Meta has responded to these concerns by highlighting the continuous improvement of their models to mitigate unintended responses. Despite these efforts, the integration of AI chatbots into social media is feared to deepen existing controversies surrounding the handling of sensitive content. As Meta proceeds with this integration, the potential for AI-induced misinformation and bias on social media platforms remains a significant concern, prompting calls for careful consideration of AI’s role in sensitive content domains.
