AI is a great 20% productivity booster - but 20% and 20x are orders of magnitude apart.

I like this one for image models: "Give me an image where every pixel is the exact same shade of blue."

Seems straightforward, but I'm yet to have a model succeed.

Data engineering feels like the DevOps of AI/ML. Important, but you have to have the right interests for it.

This comes from someone doing web dev after studying data science, so I don't know how well it reflects reality as I've never worked as a Data Eng. myself

> Data engineering feels like the DevOps of AI/ML.

I wouldn't label it that way, AI/ML has its own ops and devops which is very distinct from data engineering.

Feels like an opportunity to me - when big tech is whiffing small builders have a chance to fill the gap