Building trustworthy AI: The role of open source in development

Watch this video to learn more about the future of GenAI with insights and tips from the CEO of CrossComm.

Don Shin, founder and CEO of CrossComm, sits down with the All Things open team to talk about the current and future state of generative AI. Don discusses two key aspects: technical evolution and commercial adoption of AI. He notes that while generative AI is advancing rapidly, there is still significant room for improvement, particularly with large language models (LLMs). These models, trained on vast text datasets, have become increasingly sophisticated, enabling them to generate intelligent responses across various media types.

Key takeaways

• Rapid evolution of generative AI presents both technological and commercial opportunities
• Importance of large language models (LLMs) and their capabilities
• Open source frameworks enhance security and trustworthiness in AI applications
• Careful evaluation of third-party LLM services is essential for compliance and data privacy
• Experimentation with generalist vs. specialized models can lead to better outcomes and cost savings

Don emphasizes the importance of an open source approach to developing AI applications. Open source models are more secure, reliable, and trustworthy, which is crucial as AI becomes integral to business and societal infrastructure. He advises organizations to carefully evaluate third-party LLM services, focusing on terms of service and data privacy.

Additionally, Don shares best practices for leveraging generative AI. He recommends starting with generalist models, which are often more powerful than specialized ones, and encourages experimentation with different model types to find cost-effective solutions. By prioritizing open source practices and understanding the nuances of AI technologies, developers can better navigate this evolving landscape.

More from We Love Open Source

• Getting started with Llamafile tutorial
• How Netflix uses an innovative approach to technical debt
• Evolving DevOps with productivity and improving the developer experience
• Harness the power of large language models part 1: Getting started with Ollama