Why Samsung top AI executive thinks incrementalism is hobbling AI
summary
Argues that incrementalism is hobbling AI and shows why fixing isolated ML components (ASR, TTS, intent handlers) misses the hard problem: the interconnections and chain-of-thought. Uses voice assistants as example of an audio→ML pipeline, cites global hand-tool sales at $38 billion and smartphone market north of $409 billion to contrast helpful vs transformational products. Calls for model-centric research plus data observation and human intuition to build controllable, integrated systems that avoid the uncanny valley.
AI-written summary · about
chapters (7)
- 0:00Introduction and speaker intro
- 1:46Defining transformational AI
- 3:30AI's history and incrementalism problem
- 6:43Voice assistants and current AI limitations
- 9:52Uncanny valley and AI expectations
- 10:13Research approaches in AI development
- 14:05Key takeaways and call to action
related talks
11mSuccessful AI adoption
10mKeynote: The Last Mile Problem: Why AI W...
46mAI and ML, separating hype from what actually matters
24mA Call for Human-Centered AI
04mInterview
45mWill Skynet sort you out? An outsider's view on how to stay relevant in the era of GenAI
29mSteering Your Company Through the AI Revolution
15mon the AI Dilemma