I tried to look up what the author is actually developing with the help of AI but it seems like his primary output is hand-wavy articles breathlessly hyping up AI and urging programmers to use AI for everything.

Running gcc or go build doesn't burn incredible GPU power.

I also have gcc or go on my machine.

Proprietary language models are the proprietary compilers or languages of old.

As the blog post suggests, let's learn the lessons of the past.

If I can't run a tool locally (that includes 'compiling' the model from data) , it is a liability not a tool.

Even local models are like requiring proprietary firmware blobs as a programming tool.

Useful or not, this situation is not desirable.

Yours, a local coding assistant user (ollama and friends)

"programming was done at the most granular level"

My understanding, though it may be wrong, is that early on, "programmers" were people who typed in the work of others with little decision making.