Keep Calm and Data On
summary
Crisp, measurable claims: data‑center emissions hit about 3% of global emissions (2024 estimate), US private AI investment tops $109B in 2024 versus $9.3B (China) and $4.5B (UK), and small models cost ~0.06 W per 1,000 queries while flagship LLMs use on the order of 1,000× more per query. Practical cautions follow: ops‑first not POC‑first, ROI attribution is messy across pipelines, guard against inflated LLM reasoning and hidden deployment costs, and keep human validation and domain expertise at the center.
AI-written summary · about
chapters (25)
- 0:00Introduction
- 1:07AI impact panic?
- 2:11Data science since 2014
- 6:28AI inflection point
- 9:21AI-first companies hype
- 11:28Generative AI costs
- 15:05AI slop truth problem
- 16:51AI regulations policy
- 18:39AI colonialism biases
- 21:31AI literacy amateur experts
- 23:44LLMs fail reasoning
- 25:31DataOps instant gratification
- 26:56Human validation AI
- 28:40Trained data scarcity
- 30:04GenAI ops culture
- 31:06AI DIY professional disruption
- 32:29AI navigation principles
- 34:13AI spend ROI struggles
- 36:39Avoiding POC trap
- 38:27Focus needs avoid GenAI
- 39:31Data scientists role evolving
- 40:58AI fluency regulatory focus
- 42:02Org misalignment priorities
- 43:29Process vs outcomes
- 44:31Conclusion human authenticity
related talks
21mThe AI Iceberg: Navigating Hidden Costs to Reach Scalable ROI
34mThriving in an AI era 🔮
54mGradient ascent for your career
41mAn Art, a Science, or Your Next Job Title?
46mPanel Discussion: Business Side of Data Science
18mSmall AI Teams with Huge Impact
56mDeveloper on a Mission
01h 16mAI Advantage Reinventing SME Operations