Keynote: Generative AI in Mobile Applications
summary
Meta releases Llama 2 and a code generation model and shows real product integrations: image and sticker generation from text, an AI assistant that searches the web for up-to-date facts, and generated Python A* pathfinding code plus tests that actually run. Models train on trillions of tokens, hit context window limits and hallucinate, so mitigations use instruction prompting, tools/plugins/retrievers and input/output validation. Claim: generative AI accelerates engineers and can auto-fill forms, schedule meetings, and reduce required screen time rather than replace jobs.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:23Paradigm Shifts
- 2:02What is Generative AI
- 3:18Generating Content
- 4:21IntentBased Computers
- 5:43Challenges
- 7:13Solutions
- 9:32Accelerator for Software Engineers
- 10:45Programming Assistant
- 12:11User Experience
- 14:28Large Language Models
- 15:44Text Completion
- 16:52Safety
- 18:53Meta
- 21:21Conclusion
related talks
32mHow generative AI is Changing the way people interact with computers
43mKeynote: Applied NLP in the age of Generative AI
58mKeynote: The past, present, and future of AI for application developers
21mRevolutionizing software creation: Harnessing the power of GenAI
11mUnlocking Potential with Generative AI - Keynote Speech
29mImpact of #GenerativeAI in Development and #SoftwareTesting
21mOpportunities of Generative AI
54mKeynote: AI-Powered App Development