4 ms·
one better approach might be to consistently introduce useless variables with very telling names.
by adgjlsfhk1 4y ago
one better approach might be to consistently introduce useless variables with very telling names.
- zirgs 4y agoAI could learn to optimise those away. Compilers already do this.
- jruohonen 4y agoHopefully AIs will do that because it is not difficult to imagine that bad actors will try to follow the same logic as commercial/political/etc. poisoners by introducing vulnerabilities to code with a hope that an LLM will pick those up in the next round of learning.