Example: AlphaFold for protein folding, bringing research to production
Other use cases: Product cataloging, customer service operations, internal support initiatives
Use Cases in Companies
Vodafone (Ignacio Garcia):
Global director of data analytics and AI
Use cases: Customer call summarization, predictive maintenance, network deployment
Adoption of Google Cloud for secure, scalable deployments
Blue Core (Arvind Krishna):
Head of engineering, data science
Use cases: Personalized marketing, better model performance with Gen AI
Problem: Standardizing product data taxonomy using generative AI for improved model accuracy
Technical Design Approaches
Kevin (Google Cloud):
Focus on internal/external data integration
Use of embedding models and vector databases (e.g., matching engine, PG Vector)
Application in non-text/image data analysis (e.g., Full Story's sequence model)
Stack for production: Vertex pipelines, metadata, high compute nodes
Technical Considerations in Deployment
Ignacio (Vodafone):
Focus on scalability, data security, and compliance (GDPR in Europe)
Integration with Google Cloud’s Vertex AI for rapid deployment across different languages and regions
Arvind (Blue Core):
Technical approach to product categorization problem using Gecko embeddings and text bison prompt
Initial attempts using GPT models; better success with Google’s Palm 2
Business Value Realization
Donna (Google Cloud):
Business results from AlphaFold (experimental acceleration, drug discovery)
Employee productivity, intuitive experiences using generative AI
Arvind (Blue Core):
Achieved accurate data results and cost efficiency
Overcame scalability issues of traditional AI
Ignacio (Vodafone):
Tangible improvements in customer service with call summarization
Potential in chatbots for customer interaction, copilot systems for internal efficiencies, and knowledge management
Faster deployment with lower cost using Vertex AI
Learning and Advice
Emphasis on continuous learning and experimenting (Kevin, Google Cloud)
Encourage experimentation and safe environments for trying new ideas (Ignacio, Vodafone)
Aligning with practical short-term use cases due to rapid advancements in Gen AI technology
Closing Notes
Panelists stress importance of hands-on experience, adequate investment in tools and platforms, and clear understanding of business metrics
Advice includes managing the expectations of stakeholders, fostering healthy experimentation environments, and ensuring foundational investments in data and AI infrastructure.
Summary
Generative AI holds significant transformation potential across various industries.
Hands-on experimentation, quick iterations, and robust technical foundations are key to realizing tangible business value from Generative AI implementations.