The Hello World Show
summary
Live on stage a panel collapses hype into nuts and bolts: AI is an engineering problem that spends 80–90% of time cleaning data, produces deployable artifacts — models are files (protobuf/ONNX/.pth) — and needs DevOps for retraining when models go stale; biased training yields failures like skin‑color soap dispensers or pet detectors that ignore dogs; message patterns are explained with a McDonald’s analogy showing pub/sub, durable messages, correlation IDs and a process manager; geoengineering and carbon removal numbers are proposed (25k t SO2 ≈ 0.5°C; Carbon Engineering < $100/t CO2).
AI-written summary · about
chapters (9)
- 0:00Welcome
- 1:26Panelists
- 2:35Seth Juarez: AI and engineering
- 14:58Tess Ferrandez: Biased AI experiences
- 26:29Jimmy Bogard: Messaging patterns
- 36:25Richard Campbell: Climate change and AI
- 48:35Imposter syndrome
- 50:22Real value of advanced AI
- 56:26Closing
related talks
23mLive Fireside Chat
34mKeynote: Lighthearted Thoughts on Conversing About Data
01h 24mIt’s People! Code is made out of people!
34mThe Human-in-the-Loop is Tired
26mtalks about being a good AI Coach
34mFinding a Way Out
05mOpening Remarks
25mNo Code, No Problem: Building Powerful AI Agents with Just Your Words