From theory to practice: The rise of Generative AI
summary
Generative AI shifts from novelty to infrastructure: RAG delays enterprises building private models by 3–5 years by letting LLMs ingest company data at query time; DALL·E, Whisper and Suno put image, audio and music generation into cheap workflows; GPT‑4o and 4o‑mini make multimodal, faster foundation models practical; Nvidia GPUs (A100/H100) remain the bottleneck with 12–18 month lead times causing product delays; cloud ownership consolidates around Google, Microsoft, Amazon and Anthropic; autonomous agents are hyped but years from full production.
AI-written summary · about
chapters (24)
- 0:00Introduction and speaker background
- 1:08AI's inevitability and embracing change
- 2:12The generative AI revolution begins
- 3:16Classical vs generative AI
- 5:21RAG and its impact on AI adoption
- 6:27Autumn 2023 breakthroughs
- 8:14Cloud giants and AI dominance
- 10:42The five major AI players
- 12:27Open models and their limitations
- 14:15Winter 2024: Microsoft's Sunno
- 16:50Misconceptions about AI memory
- 19:21Spring 2024: Nvidia's dominance
- 21:50AI certifications and industry shifts
- 22:33Nvidia's chip shortage and impact
- 25:25Summer 2024: Multimodal models
- 27:15Model selection and costs
- 28:16EU AI Act and job concerns
- 31:06Autumn 2024: ChatGPT advancements
- 32:18Understanding AI's unpredictability
- 33:21AI hype cycle and reality
- 35:492025 predictions: Autonomous agents
- 36:51Building autonomous AI agents
- 40:42Summary and call to action
- 41:28Q&A session
related talks
01h 14mLatest trends in AI
49mGenerative AI: Concepts, Tools, and Applications for Data Professionals
22mGenerative AI & its Transformation of Business, Research, and Society
17mGenerative AI: Keeping it real
41mThe AI boom: what and why?
37mGenerative AI: Beyond (Just) Hype
52mBeyond the AI Hype: What's Real, What's Next
22mGenerative AI: Opportunities for the New Age of AI