The Meta made significant strides in artificial intelligence, launching two smaller versions of Llama 3, a new large language model, and enhancing the virtual assistant Meta AI. This move is part of Meta's strategy to cover the distance that separates it from the OpenAI, a leading company in the field of artificial intelligence.
Llama 3: Creativity and intelligence
The new Llama 3 models, with 8 and 70 billion parameters respectively, are integrated into the virtual assistant Meta AI, which is expected to play a more important role in Facebook, Instagram, WhatsApp and Messenger. Meta's goal is to compete with Chat GPT, the popular artificial intelligence model developed by OpenAI with its support Microsoft.
Llama 3 has new coding features and has been trained using images and text. Its next versions, expected in the coming months, will add more advanced features, such as the ability to compose larger designs with multiple steps.
Collaborations and developments
Meta also announced a new partnership with Alphabet, the parent company of Google, to include real-time search results in the Assistant's responses. At the same time, the Meta AI assistant is going to be made available in new markets such as Australia, Canada, Singapore, Nigeria and Pakistan.
Meta is “still looking for the right way to bring this service to Europe”, as the legislation of the European Union is stricter in matters of privacy protection.
Requirements and challenges
The development of artificial intelligence brings significant challenges to the fore. Meta claims to have mitigated the issues Llama 2 faced by using "higher quality data" and "synthetic" data generated by artificial intelligence.
At the same time, the company faces criticism for the ethical dimension of artificial intelligence. App developers have complained about Llama 2's weaknesses, while Google suspended its Gemini AI imaging feature after the model was criticized for showing inaccurate depictions of historical figures.