Navigating NLP: From Prototypes to Production

“The last nine months, no matter what we want to talk about, our customers want to talk about one thing, which is Gen AI.”

One year ago, ChatGPT woke the entire world up to the possibilities of Generative AI. Since then, the conversation around AI has not ceased, with constant questions being raised about its safety, accuracy, and potential implications for the tech world.

In this episode, Dan hosts a panel at NYC Tech Week 2023, discussing the evolution and potential of Generative AI with Jacopo Tagliabue of Bauplan, Cat Miller of Flatiron Health, Raghvender Arni of AWS Industries, and Justin Zhao of Predibase. Together, they discuss the journey in taking Generative AI from prototype to production, delving into the need for a stable foundation of best practices in data and software development, how to overcome the challenges of evaluation and performance, and the cruciality of red teaming and guardrails to ensure the viability and usability of AI models.

Timestamps:

• 00:00 - Intro
• 02:19 - About the guests
• 12:58 - GOFAI vs GenAI: Should you just be using a classifier?
• 18:54 - AI + health data
• 23:24 - How do you make sure evaluation works?
• 24:57 - The foundations needed to take GenAI to the next level
• 34:22 - Is now the time for experimentation or business models?
• 36:49 - The science fiction of today that will be commonplace tomorrow
• 42:42 - Outro