Another problem is that we're turning any problem into a black-box, which takes the fun out of problem-solving.
It’s cheaper to move from one model to another than it is to train a general purpose model yourself (to say nothing of domain-specific smaller models or anything open source.)
I’m not sure about problems turning into black boxes, LLMs are pretty explicit in my experience when producing a solution (good or bad.) _How_ they came about that solution _is_ a black box, but that’s not a new problem.
We can use OpenRouter to build agents with any LLM and switching your agent to a new model is a one-line code change. We can write MCP tools that work with most of the decent models.
Honestly, I think we may be entering a period where things start to decentralize and money starts to move towards startups building interesting tools and agent workflows instead of a handful of giant companies training frontier models.