Meta, the company behind Facebook and Instagram, has stepped up its AI game by releasing its LLaMA series of language models and launching a ChatGPT-like service. These models are now available as largely free-to-use, customizable, and modifiable software, setting Meta apart from competitors like OpenAI and Google. Mark Zuckerberg aims to integrate AI into all Meta services, positioning the company as a major player in the AI space.
The newly introduced Meta AI Assistant, powered by the latest LLaMA-3 model, will be integrated across key Meta products such as Instagram, Facebook, WhatsApp, and the Facebook Messenger. Additionally, Meta has launched Meta.ai, a dedicated website for the chatbot, inspired by ChatGPT and Perplexity search engine, providing access to a proprietary image generator.
Mark Zuckerberg notes that Meta AI has the potential to become a significant product and emphasizes Meta’s commitment to providing the most intelligent AI assistant that people around the world can freely access, without requiring a subscription.
Unveiling two smaller versions of the LLaMA-3 model for the open-source community, Meta has already garnered praise for the models’ remarkable knowledge and ability to generate targeted computer code and handle complex tasks. LLaMA 3 is expected to be available as a service through various cloud providers.
However, the use of training data for such models has sparked controversies, with Meta facing challenges in collecting sufficient training data for LLaMA 3. While the specifics of the training data remain undisclosed, Meta has hinted at employing “public” internet data and synthetic data for training.
🤔 What are your thoughts on companies using publicly available internet data and synthetic data for training AI models? #AIethics