Discussion
they have not gone out and said that i need to use AI to keep my job
but they also didn't explicitly say it at the last place before i got laid off either
this is day 2 since The Meeting
yesterday, i used Claude. i already see how folks want to use it instead of just using internet searches and how easy that would be
i still want to manually confirm anything it tells me. hold me to this lowest of fucking bars, please
didn't have to use claude yesterday. probably gonna ask claude to build an identity app in rails over the weekend to burn tokens since that's what they want me to do
i've cracked it
i figured out why folks are so crazy over generative AI
these nerds want someone that is always ready and willing to listen to every stupid idiotic idea they have and be the sycophant they think they desire
i'm sitting here after using claude a moderate amount yesterday and a voice in my head says "hey we should pitch x idea to claude"
@hex I tried to get Gemini's help coming up with a stack computer architecture (not metered at work so I'm pretty sure they wouldn't care it wasn't directly work related) and it took me three tries to even get it to realize that you can't increment the program counter at the end of the instruction because that would result in a clock cycle spent doing nothing before starting the next instruction. The first time it was like "Oh yeah I'll just use a lookahead adder!"
So not only are LLMs sycophantic, they also confidently give replies about things they know nothing about.
@hex I'm guessing people will experience the Gell-Mann Amnesia effect with LLMs just like they do with journalists, where even though they know LLMs can be confidently wrong about things they themselves are experts on, they'll continue to trust them on things they know nothing about.
Install bonfire.cafe
Get the full app experience
