Can AI replace programmers?
summary
Claim: AI can generate working code but often at high cost and low readability. Examples: genetic programming evolves a Python fizzbuzz that required days of parameter tuning, eight upfront tests and stalled around five of eight cases before eventually succeeding; a Java code generator produces brittle, unreadable scaffolding; AlphaGo-style self-play solves Go but needs huge data and compute. Notes on tools and risks: clang / LLVM and CI pipelines, deep learning training uses lots of electricity; human-in-the-loop and careful tests remain essential.
AI-written summary · about
chapters (12)
- 0:00Start
- 4:26Ascribing Mental Qualities to Machines
- 12:16Automate everything?
- 16:37Augment everything
- 19:43Automation is a Good Thing
- 22:04Automation has a carbon footprint
- 23:59Automation is magic
- 40:09Can Al replace people?
- 45:21Genetic algorithms
- 46:53What is a genetic algorithm?
- 52:43DEAP
- 55:35Writing the tests is hard
related talks
20mWill AI Replace Coding or Improve It?
39mBack to Basics: Cutting Through the AI-Generated Code Noise
43mWill AI kill the developers job?
14mWill AI replace All coders?
33mAI assisted software delivery
44mAI Code Generation - Benefits, Risks and Mitigation Controls
54mSoon replaced by AI? An overview of code generation
41mAI or Not AI? We Don't (Yet) Believe the AI That Replaces Devs!