Unlocking the Potential of AI - April Edwards & Henk Boelman - NDC Porto 2023

Ai

AI has evolved from proof-of-concept to generating next tokens, predicting contexts, and practical applications.

Key takeaways
  • AI has come a long way, from being limited to proving up on existing data to now learning from it.
  • Generative AI can predict the next likely token, given the context.
  • Custom models can be built with Azure OpenAI, allowing for more control over the response.
  • Error handling is important, and safety mechanisms can be implemented to prevent undesired behavior.
  • AI models can be used for search applications, and Azure cognitive search can combine multiple sources.
  • Vectors and factors can be used to search for similar documents and contexts.
  • The future of AI models is flexible, usable, and adaptable to different problem spaces.
  • The potential of AI is vast, but it is important to consider the impact of AI on individuals and society.
  • Collaboration between humans and AI is key to harnessing the potential of AI.
  • GitHub Copilot chat can be used to explain code and add comments, and can even generate unit tests.
  • The indexing process can be managed through Azure OpenAI, allowing for efficient storage and retrieval of data.
  • Factoring and vectorization can be used to improve the speed and accuracy of search results.
  • Microsoft is building custom models for Office, demonstrating the potential of AI in practical applications.