Towards “Thinking Machines”
summary
Covers progress toward machines that understand, learn, reason and create by drawing on advances in machine learning, natural language understanding, speech tech, computer vision and deep learning. Explores hypotheses and algorithmic ideas for unsupervised learning of word senses without explicit grammar, extracting and generating knowledge from language, reasoning over that knowledge to solve problems and innovate, and applying these components to improve machine translation and conversational systems.
AI-written summary · about
chapters (25)
- 0:00Intro
- 2:42We have come a long way...
- 4:29Today Machines can...
- 5:51But can they...
- 11:00How do we get here?
- 12:45The deeper questions: What is...
- 15:08Composition / Segmentation!
- 18:09Semantics / Word Similarity!
- 22:09Semantics / Sentence Similarity!
- 23:38Disambiguation!
- 26:57What senses "make sense" together
- 27:50Sense of a word | Sense of context
- 31:54Putting it all together!
- 35:09An Equation of Thinking!
- 37:16Is this how we do it?
- 40:06Deep Learning approach
- 42:32An Architecture for All
- 44:33What is the atomic unit of LANGUAGE?
- 46:05What is the atomic unit of KNOWLEDGE
- 49:40Language → Knowledge
- 51:03Knowledge → Language
- 52:08Key is Synthesis & Analysis
- 55:12Three Knowledge representations
- 56:21Towards Thinking Machines!
- 58:22Two questions to keep you up!
related talks
43mAI Systems: Foundations for Artificial Minds
59mIntelligence and Machines: Creating Intelligent Machines By Modeling the Brain
34mCausality, robustness, and natural language understanding in ML
04mInterview
21mLeveraging AI to enhance human creativity
24mMachine Intelligence gets real
03mInterview
04mFrom big data to human-level artificial intelligence