Generative AI: Path to Value in practice
summary
Generative AI projects show a surprisingly high failure rate at every stage because teams lack shared success criteria, pick the wrong models or DIY tooling, and treat proofs as production-ready architectures; common blockers are brittle PC architectures, curated data slices that break at scale, missing operations/security signoff, and poor adoption. AWS points to using managed services, Bedrock agents and Nova models for better price‑performance and indemnified output, running prototypes with the center of acceleration, instrumenting adoption, and prioritizing data‑ready use cases to turn POVs into measurable value.
AI-written summary · about
chapters (7)
- 0:00Introduction to Gen AI value
- 1:06Challenges in Gen AI value extraction
- 7:51Moving from proof of concept to production
- 13:51Risks of trial and error in AI
- 19:29Why adoption doesn’t happen automatically
- 25:11Measuring and realizing AI value
- 27:40Summary and key takeaways
related talks
17mGenerative AI: Keeping it real
27mGenerative AI for technology leaders
20mBeyond the Buzz: Discovering High-Impact Use Cases in the Generative AI Era
23mGenerative AI use cases
18mNavigating the growth frontiers of generative AI
22mGenerative AI - A Pragmatic Business View
22mGenerative AI & its Transformation of Business, Research, and Society
47mLessons learned from using generative AI dev tools