@squinky so, how do you feel about your job and your self-esteem?
I have found that LLMs are great at doing some of the things I really hate, like looking up API references. Also, they're just great at code review and path analysis. Finding the cause of a bug can be really quick.
Also, doing search-and-replace work that doesn't match regexes easily. "Change every place we call this constructor to include the new required parameter." A real pleasure.
@squinky It's like if the Olympics measured how much Wheaties you eat and ignored your 100m sprint time.
@squinky I think the numbers on agentic AI are not in yet.
@squinky Microsoft Research showed about 55% productivity gains. Which is impressive but not 10x. This is 3 years ago, though; a millennium in AI time.
@evan or remainder of the Usenet culture it stems from 🤷
@squinky that's extremely confident in the value of the tools!
My latest books cartoon for @theguardian.com@bsky.brid.gy #manosphere
@squinky so, that's pretty interesting.
My main question is about the employer warning. Is it that you have to hit some productivity milestones like velocity or defect rate, which you couldn't possibly do without LLMs, or is it literally that you have to use LLMs regardless of their benefits?
@oivaeskola hahahahaha