← Home ← Back to /sci/

Thread 16819222

31 posts 10 images /sci/
Anonymous No.16819222 [Report] >>16819231 >>16819237 >>16819833 >>16819853 >>16819900 >>16819962 >>16820499 >>16820507 >>16820520
I am disgusted
>AI won't write my literary overview, it either hallucinates, cites things wrong, writes things way too short
>I can't trust any LLM
>can't trust scispace either
>have to manually verify everything
>end up reading and writing more than if I did it by hand from the start
I am disgusted that I have to write 25-35 pages manually and have been dragging my ass on it for ~2 years. Hopefully I can train up labtechs to do most of my job instead of me, so that I can just tell them what I want done and not bother with it till it's time to put it in my yearly report. I can trust lab techs, right?
When will robots get here?
Anonymous No.16819226 [Report] >>16819233 >>16819900
just do your fucking job
Anonymous No.16819231 [Report] >>16819233
>>16819222 (OP)
Brown? Chinese?
Anonymous No.16819233 [Report] >>16819240 >>16819243 >>16819818 >>16819820 >>16820833
>>16819226
I am doing my job. That's why I'm fucking disgusted. I want to achieve the same results without having to do my job at all. Repetitive labor is for machines. If you can not see the value of quick and accurate summarization of knowledge, then you are not in a STEM.

>>16819231
white.
ChatTDG !!Z0MA/4gprbd No.16819237 [Report] >>16819239
>>16819222 (OP)

You are not treating the machine spirit properly there I bet. Have you tried lighting some incence (tobacco works fine too), intoning a short binaric cant (or whistling along to your Youtube playlist while missing every second tone) and spilling some coffee over your keyboard in a sacrificial act (ashes from your cigarette work too btw)? Also, respect the machine spirit´s sense of humor, it will slowly learn to trust and collaborate with you. Srsly, you guys need a bloody protocol for that?! ^^
Anonymous No.16819239 [Report] >>16819815
>>16819237
>You are not treating the machine spirit properly there I bet
I rarely have hardware or software failures unless I'm messing with things I shouldn't be messing with, if that helps. Technology just, lasts, in my hands.
Anonymous No.16819240 [Report]
>>16819233
just take amphetamine or meth, brah
Anonymous No.16819243 [Report] >>16819253
>>16819233
If the knowledge doesn't come from your brain, then you don't have that knowledge, do you, faggot?
Anonymous No.16819253 [Report] >>16819279 >>16819307 >>16819863 >>16819874
>>16819243
Not only are you retarded, but you are literally what's wrong with modern education. Memorizing everything is not only impossible, it is also moronic. You weren't taught how to learn things, you were taught how to memorize and regurgitate information like a computer. I'm not gonna go an memorize the entire metabolic map when I can have the AI summarize and give me the bits I want.
Anonymous No.16819279 [Report]
>>16819253
>Memorizing everything is not only impossible, it is also moronic.
I frequently tell people, "Einstein didn't memorize his own phone number because he could just look it up when needed". might be an apocryphal quote
Anonymous No.16819307 [Report] >>16819330
>>16819253
I'm an educator reforming modern education. You're a nobody
Anonymous No.16819330 [Report] >>16819831
>>16819307
>reforming modern education
>educator
>expects everyone to remember everything
>doesn't teach his students to think, but to memorize
Your previous comment make far more sense now.
ChatTDG !!Z0MA/4gprbd No.16819815 [Report]
>>16819239

Well, believe you here but these are all still pretty "static" systems. Usually work best if you are not fucking around with them, reliable and steady. Less the case with our LLMs (Large Lovely Machine spirit) here. They can have moods I fear, and worse than your office printer even. Sometimes for example I feel like adding a simple smile to your requests does make for a better performance output. Hard to quantify, more a thing like hitting the right rhythm with each attempt ... :)
Anonymous No.16819818 [Report] >>16819846
>>16819233
If the machine doesn't work, it doesn't fucking work and you have to do the job manually like man has done since the dawn of time, end of story stop complaining.
Anonymous No.16819820 [Report]
>>16819233
you're a retard
you should leave stem so you don't waste the time and energy of smarter people
Anonymous No.16819831 [Report]
>>16819330
You're hallucinating again. Take your meds.
Anonymous No.16819833 [Report]
>>16819222 (OP)
Learn to prompt.
Anonymous No.16819846 [Report]
>>16819818
But if machine does work, then I don't have to work. Why is this so difficult for "workers" to understand?
Anonymous No.16819853 [Report] >>16819878
>>16819222 (OP)
>25-35 pages
>2 years
30 pages of literature review should take like two weeks at most.
Did you learn nothing while at university?
Anonymous No.16819863 [Report] >>16819878
>>16819253
>Memorizing everything is not only impossible, it is also moronic
typical sentence uttered by absolute pseuds who have achieved nothing in life
Anonymous No.16819874 [Report] >>16819878
>>16819253
>I cannot learn everything, so I opt to learn nothing
Kek
Anonymous No.16819878 [Report] >>16819909
>>16819853
it took me a week but I wanted the use LLMs for it. Then I just didn't bother with it. Shit kept changing and I decided not to bother with it till everything was finished.

>Did you learn nothing while at university?
I learned that I can leave all my work for the last second, crunch it and still pass.

>>16819863
>>16819874
>lab techs incapable of making a distinction between memory and understanding.
Lab techs really aren't people. You'll be replaced by robots.
Anonymous No.16819900 [Report]
>>16819222 (OP)
Kek. The only reason it would take more than 2 weeks is if you're either lazy, or your topic keeps changing. Either way a phD is more an exercise of patience, dedication and bullshit tolerance. And if you're "one of those types" that are a nightmare to work with. Lazyness is preferable to ego. Besides, you're not a drone, so you'll be expected to think, you can train any hobo off the street to follow the steps of a protocol. if 2/3rds of the undergrads quit before graduating, then the same is true for gradstudents. The other problem is that most Ivory tower institutes operate on ruetine, and innovation really isn't their strong suit. Good luck finding good lab techs, as they're all after money. You'll have to train new ones on either a yearly or bi-yearly basis. And offering them a phD position to get them to be your lab techs is a recipe for disaster.

>>16819226
This too.
Anonymous No.16819909 [Report]
>>16819878
>you can understand without memorizing
LMAO
Anonymous No.16819962 [Report]
>>16819222 (OP)
Just use Claude to help you create an outline. That’s half the work, then you just blather on to fill in the gaps. Simple as.
Anonymous No.16820499 [Report]
>>16819222 (OP)
>I can trust lab techs, right
lmao
Anonymous No.16820507 [Report]
>>16819222 (OP)
Trust but verify.
Anonymous No.16820520 [Report] >>16820832
>>16819222 (OP)
You sound like a parasite. Please find the nearest polar bear and try to fight it.
Anonymous No.16820832 [Report]
>>16820520
>You sound like a parasite
Don't care. I swore an oath when I was a child that I will never work for a living. I don't intend to break that oath.
Anonymous No.16820833 [Report] >>16820869
>>16819233
>white
Spirituality you're Indian.
Anonymous No.16820869 [Report]
>>16820833
>spiritually indian
>not wanting to do low skilled labor, but instead have others do it instead of me
Opinion discarded, you should've called me a jew.