In other news, OpenAI is launching a mini version of its latest AI model, designed to be faster and more affordable. Meanwhile, Reddit is changing its terms of service to prevent AI companies from scraping its data without payment. There are also concerns about AI models potentially becoming less intelligent and even collapsing. Finally, LinkedIn is offering AI-written cover letters, and YouTube is making it easier to report and take down AI deepfakes.
Key takeaways:
- Meta's new Llama 3.1 405B model will soon allow users to ask the AI to create avatars of them in various roles, a feature called "imagine me".
- Unlike viral avatar apps, Meta isn't charging for this feature, leading to predictions of a surge in AI-generated selfies.
- There are concerns that AI models could potentially get dumber and even collapse, despite recent breakthroughs.
- OpenAI is launching a mini version of its latest AI model, intended to be faster and more affordable than larger models.