AI vs ML
Purists love to argue the difference. For a decade, it didn’t matter much. Most AI was really ML1. Now it matters again, especially when comparing ML engineers (the older title) with AI engineers (the newer one). What they do, and what they need to know, diverge.
ML engineers read papers, adapt ideas into production systems to make an impact. That meant going deep: into pipelines and systems, yes, but also into losses and gradients. The craft combined engineering discipline with mathematical intuition.
AI engineers are newer to the game. They start from pre-trained models, usually LLMs, and engineer intelligent behavior out of them: prompting, retrieval, evaluation, and context design.
The difference mirrors the terms themselves. ML engineers train learning systems. AI engineers orchestrate them. The first group trains models; the second uses them2.
For clarity: AI is broader. ML is the subset that learns from data.
Can AI engineers train models? Of course. When they do, they become ML engineers.