Fallacies of GenAI by Speaker 3
- 1. The fallacies of Generative AI Opportunities & Challenges in productionizing
- 2. Introduction Who am I? Overview of Session Goals • Fallacies of Generative AI • Opportunities & Challenges • Strategies for product • Impacts and Future of AI Agenda
- 3. The AI Pyramid Don't care about AI Use AI at work Integrate AI in Enterprise solutions Build AI from scratch Big tech landscape Opportunity Opportunity
- 5. Hype Cycle Trends that would reach productivity < 2 years • Retrieval Augmented Generation (RAG) • GenAI Enabled applications • GenAI Workload accelerators • GenAI enabled virtual assistants
- 6. Common Fallacies š Fallacy 1: Generative AI is a Magic Bullet š Myth: One magic pill for all our problems š Reality: Needs specific training and fine-tuning, richness of AI comes from your data š Fallacy 2: Generative AI Understands Context š Myth: Perfect comprehension of context and nuance š Reality: Limitations in understanding complex, nuanced queries š Fallacy 3: Generative AI is Completely Autonomous š Myth: No human intervention needed š Reality: Requires human oversight and validation
- 7. Data Powers AI - Challenges The data needed to power AI are scattered Data are not NLP friendly Solution: Transform your data platform as single source of truth with NLP friendly JSON schemas Data platforms are not built for AI AI may not be able to chew our complex production schemas Solution: Keep the data for AI mostly in denormalized form like a datawarehouse Complex production data Duplication of data results in Bias Solution: Qualitative assessment of data Data duplication
- 8. Challenges in Productionizing Bias and Hallucination Explanation of bias in AI models Examples of hallucination and its impact Inference and Infrastructure Technical challenges in deploying Generative AI Infrastructure requirements and associated costs Cost Considerations Budgeting for AI development and maintenance Cost vs. benefit analysis
- 9. Product Strategy š Not all products and solution need Generative AI š A successful Generative AI based product exhibits o Generative AI is not a force fit into the product o Product delivers value even without Generative AI o Solves a real user pain point or a problem o Use of generative AI amplifies the value delivered o Automates a complex set of operations within the product
- 10. Idea to Product PoC / MVP with commercial APIs (ChatGPT, Gemini) Rich business domain dataset Self hosted LLM models A/B & RC with feedback Fine tuning 01 02 03 04
- 11. Model Landscape
- 12. Team Composition AI architect Data Scientist Data Engineer Developer Business Analyst / Scrum Master Quality Analyst Product Manager NLP Engineer (Optional) MLOps Engineer (Optional)
- 13. Other Considerations š AI assisted software delivery š AI assisted content creation š AGI? š AI and workforce š Success Stories š Adobe – Generative Fill š Microsoft 360 – GenAI Designer
- 14. Recap š Opportunities and challenges š Your AI Strategy
- 15. Questions?