pedeypops · HackerTrans

6개월 전discuss

I agree. I've seen people insist moving to a newer model or fine tuning will make the output more clever, "trust me", sometimes without providing any evidence of before and after for the specific use case. One LLM project I saw released was prettymuch useless, but it wasn't the use case or the architectural limitations that were the problem, nope the next thing on the roadmap was "fixing" it by plugging in a better LLM.

pedeypops

2년 전discuss

"the workers on the project are not, as a whole, sufficiently experienced or well trained"

Plenty of executives pretend that if you throw enough procedures and micromanagement at cheap entry-level employees then they can produce similar stuff as experts, and this seems to be evidence Boeing is trying and failing to do just that. Any executive can put procedures in place, and good training can do great things, but it's hard to say you care about quality then slash pay for good labor. Take this line from the actual report:

"Michoud officials stated that it has been difficult to attract and retain a contractor workforce with aerospace manufacturing experience in part due to Michoud’s geographical location in New Orleans, Louisiana, and lower employee compensation relative to other aerospace competitors... it is too early to determine if the new training alone will result in a notable decrease in nonconformances and CARs issued. According to a NASA official, further quality assurance challenges related to the workforce stem from work instructions that lack explicit details on how to perform the task and with what tools. Some technicians reported they had to hunt through layers of documentation to identify required instructions"

comments