LLMs Will Replace 8-Track Duplication Engineers
Summary
The piece analyzes an NP-hard 8-track partitioning problem and uses it to benchmark human vs algorithm vs LLM performance. It shows that humans historically beat standard algorithms like LPT and KK, but modern LLMs can approach or surpass human performance under careful prompting. It discusses data sources, methodology, and the role of AI benchmarking in complex optimization tasks.