In April 2024, Meta released their newest generative AI model, Llama 3. This new release builds on the success of Meta’s Llama-class models, which have been downloaded over 170 million times to date.
Along with being a powerful model, Llama 3 is open-source, meaning that anyone can use or modify the model for their own needs. In two recent articles, Meta shared how several organizations are using this new large language model to help students, patients, and customers.
Using Generative AI for Math Education
With their capacity to answer questions and dynamically adapt to their user, generative AI tools like the Llama large language model offers many great use cases across education.
South Korean ed-tech start-up Mathspresso worked with Quanda.ai used Llama 2 to develop MathGPT – a large language model focused on math education. The goal was to develop a math-focused LLM that could be adapted to local factors, like curriculum, exams, teaching style, and more. In the article, Mathspresso CEO Jake Yongjae notes, “Commercial LLMs like ChatGPT lack the customization needed for the complex education landscape”. Open-source LLMs like Llama can help fill that gap.
Using Generative AI to Promote Health Equity
A major challenge to health equity around the world is the ability to diagnose patients in underserved or remote emergency scenarios. Researchers with Yale School of Medicine and EPFL’s School of Computer and Communication Sciences developed Meditron, “a suite of large multimodal foundation models designed to assist with clinical decision-making and diagnosis”. Meditron uses the Llama model as a foundation, fine-tuned on medical information.
The researchers are now working with organizations like the Red Cross to share this with healthcare workers where needed. It has since been downloaded over 30k times, and is undergoing validation by doctors around the world.
Yale professor Mary-Anne Hartley notes in the article, “When applied to the medical domain, they [generative AI models] have the potential to provide life-saving advice and guidance. Yet the lowest-resource settings have the most to gain and remain the least represented.” Meditron is a positive step toward resolving this.
Using Generative AI to Make Meetings Better
Videoconferencing company Zoom uses Llama 2 models to power their GenAI assistant, Zoom AI Companion. This feature can do a number of tasks like:
- Summarize meetings, including highlights and tasks
- Catch you up if you join a meeting late
- Organize meeting recordings into chapters
- Provide presentation feedback for future meetings
Zoom is also expanding this to support more work activities surrounding meetings, such as preparing agendas, summarizing chat activity, idea generation and more.
Using Generative AI to Enhance Social Media
Naturally, Meta is also using its own generative AI models for applications across numerous platforms, including:
- A general text-based AI chat (answering questions, offering feedback, etc.)
- An agent that can help you with planning or scheduling
- Digging deeper into interesting posts
- Generating AI images as you type
- Changing styles of or animating images
The fact that Llama can be used and adapted for such a wide range of tasks underscores both how powerful LLM’s can be, as well as how important open-source AI is to the world at large.
If you’re interested in trying Llama for your business, I discussed the benefits and challenges of using open-source models in a previous post. In a future post, I’ll provide directions on how to use a tool called llamafile to launch a version of Llama 3 on your own computer. More soon!
Want to stay informed on the most important AI news in just 30 minutes a month? Sign up for the Executive Summary newsletter by AI For Business.
If you liked this post, have future posts delivered straight to your inbox by subscribing to the AI For Business newsletter. Thank you!