Same. 3.5 Sonnet seems particularly good at React and Tailwind. I'm moving away from other frameworks and libraries to React and Tailwind because of this.

I suppose various network effects have always hindered diversity, but I fear LLM training will severely hinder the adoption of new languages, frameworks, and growth of smaller technical communities. Businesses will choose to use the most automated solution rather than the best tool for the job.

Maybe the most automated solution will become the meaning of 'best tool for the job'.

The only positive I can see from this is that it should push maintainers towards stabilization faster as large API changes would result in huge productivity losses. Already true but perhaps more clear when the LLM starts producing failing tests.

Thanks for adding the specific case. I think with testing these sort of limited domain applications make sense.

It'll be much harder for more open ended world problems where the physics encountered may be rare enough in the dataset that the simulation breaks unexpectedly. For example a glass smashing into the floor. The model doesn't simulate that causally afaik