RealFeedApp Experiment on coding is not yet finished
The author conducted an experiment generating and inspecting Python code with LLMs, discovering systematic errors in logic and security.
- Field
- applying language models
- What they did
- The author conducted an experiment to test the reality of generative AI promises using code generation as a test case, employing his own scripts for analysis.
- Why it matters
- The goal is to understand how accurate the claims about AI capabilities are before publishing full conclusions and a demo video.
#llm#code inspection#python#generative ai#security
