Code + AI: Will Robots Take Our Coding Jobs? ML Applied to Programming
summary
Neural nets learn vector embeddings of code using AST paths and token contexts to predict method names, suggest variable names, and power code completion; models trained on large corpora (examples include Linux kernel data) reproduce analogies like add/remove → update and support live web demos. Deeper models use graph-based control/dataflow to learn API usage patterns and flag anomalies. A one-sheet Excel neural-net demo shows the math is linear combinations plus max; results are useful but produce surprising edge cases.
AI-written summary · about
chapters (19)
- 0:00Intro
- 1:31TIDAL WAVE
- 3:15Down The Rabbit Hole
- 5:18Classification : Images
- 7:14Translation : English
- 10:56What is Machine Learning?
- 11:39Artificial Intelligence
- 12:19Artificial Neural Networks
- 12:34Deep Learning
- 12:57How Do Neural Networks Work?
- 17:29Computationally - Dead Simple
- 22:30Learning Boundaries
- 25:07Neural Network Usage
- 25:21Neural Network Input
- 26:09Feeding to Neural Network
- 26:21Feeding Text to Neural Network
- 33:01Neural Networks for Code
- 36:59Labeling Functionality
- 41:05Other Tasks
related talks
43mWill AI kill the developers job?
43mMachine Learning for Developer Productivity
30mThe AI Revolution in Coding
32mComputer Programming Is Dead; Long Live AI-First Programming
24mNatural Language Coding Revolutionising Programming with GenAI
45mProgrammer’s Apprentice Season 2: Future Directions in AI-assisted Coding
38mWriting code live with AI
37mMachine Learning on Source Code