I've noticed this already with Claude. Claude is so good at code and technical questions... but frankly it's unimpressive at nearly anything else I have asked it to do. Anthropic would probably be better off putting all of their eggs in that one basket that they are good at.
All the more reason that the quest for AGI is a pipe dream. The future is going to be very divergent AI/LLM applications - each marketed and developed around a specific target audience, and priced respectively according to value.
In my lab, we have been struggling with automated image segmentation for years. 3 years ago, I started learning ML and the task is pretty standard, so there are a lot of solution.
In 3 months, I managed to get a working solution, which only took a lot of sweat annotating images first.
I think this is where tools like OpenCode really shine, because they unlock the potential for any user to generate a solution to their specific problem.
Apparently on Macs it's usually Command-Shift-Z?
This is all pageantry.
"I know nothing but had an idea and did some work. I have no clue whether this question has been explored or settled one way or another. But here's my new paper claiming to be an incremental improvement on... whatever the previous state of understanding was. I wouldn't know, I haven't read up on it yet. Too many papers to write."
We removed the authorship of a a former co-author on a paper I'm on because his workflow was essentially this--with AI generated text--and a not-insignificant amount of straight-up plagiarism.
Didn't even open a single one of the papers to look at them! Just said that one is not relevant without even opening it.
E.g. “cite that paper from John Doe on lorem ipsum, but make sure it’s the 2022 update article that I cited in one of my other recent articles, not the original article”
A couple of generations of students later, and these will be rare skills: information finding, actual thinking, and conveying complex information in writing.