Q&A #3
summary
A Q&A covers concrete takeaways: a working Azure OpenAI + Teams chatbot that queries internal SharePoint HR docs, advice to keep company data in Azure for security, and real experiments with survey and weather datasets; evidence of model failures includes early tests where GPT-3.5 in Google Sheets alternates between plausible output and blatant refusals, and cases where the model invents hallucinated DAX functions for Power BI; practical tools mentioned: Superpower ChatGPT, prompt stores and folder-based prompt management.
AI-written summary · about
chapters (17)
- 0:00Introduction and welcome
- 0:46Security concerns with AI access
- 2:11Philosophical discussions with AI
- 3:37Data visualization bias
- 5:43Ensuring data accuracy with AI
- 7:29Exploring historical data with AI
- 10:41AI democratizing data analytics
- 11:26Joys and frustrations with LLMs
- 13:32Data visualization challenges
- 15:20Hallucinations and AI reliability
- 17:05Managing and sharing prompts
- 20:17Prompt organization systems
- 22:46AI for debugging in programming
- 23:52AI for non-technical users
- 26:59Risks of AI-driven development
- 28:03Quick-fire tool and tip round
- 30:10Closing remarks and community
related talks
54mDeveloping Sophisticated Conversational Experiences w/ Microsoft AI Platform
46mDesigning and implementing automation and conversational AI Q&A
01h 01mUnlocking the Potential of AI
37mHarnessing OpenAI to help millions
51mThe Next Frontier: Azure OpenAI Models and the Arrival of GPT-4
48mGetting started with generative AI using Azure OpenAI Service
44mExploring ChatGPT for Improved Observability
46mAI innovation in the Microsoft Power Platform