• S&T Moderators: Skorpio | VerbalTruist

Is A.I. waiting to take over ?

whoever said the thing about emotional intelligence lacking and probably always will be is correct. Personally unless it is something I am interested in I only read the abstract, how the study was done, basic results. Expecting everyone to read everything is a little much --- as is replying as if you had read everything (to be fair) and certainly will lead to confirmation bias if you are "cherry picking" which you shouldn't be by reading the summary?
 
If the summary contains the information required for the task at hand then the summary is sufficient. Why would I waste time reading the whole book
Bc reading the whole book often leads to a more robust and complete understanding of the subject. You first would have to be in a situation where reading the book would not constitute a "waste of time". What our Schutzstaffel friend doesn't seem to get is that very often we don't have the luxury and free-time to read all the books we would like. So we do the next best thing—us not being scholars but the workforce in demanding rolls with real obstacles like deadlines, team expectations, sprint planning, load balancing, etc.—we take what we need initially. In 2025, vis-à-vis data gluttony and countless mountains of literature, the right AI model is clutch in trawling thick forests of data to serve up useful knowledge gems for what you need when you need it (as you and I know). With enough experience and time you end up reading the whole book and several others (or the equivalent of same), albeit on a need-to-know basis. Sure, in theory, you'll learn better by reading the right books front-to-back in some logical order.

However, in praxis it's all "pragmatism, not idealism." Ultimately you learn best by keeping your job and continuing to accrue hands-on real world experience. @-=SS=- doesn't understand the pros and exaggerates the cons. All of us need to test our code, anyway, so it's not some waste of time to check it line-by-line. You should lint anyway, and read every line of the code you commit first to be sure you understand what it's doing. If that's not the obvious CYA method of incorporating some LLM into your workflow successfully, then Idk what is. But LLMs do not automatically mean every code developer across the world just got lobotomized or something. And sometimes ppl who write and publish books about coding, firstly that's some ironic, slow-paced shit, and secondly, there can be a tendency for these authors to become self-indulgent and less effective at teaching. Prose quality and subject relevancy vary.

If the task at hand is your own learning, learning to learn,
That's nice and all, but this does not pay bills. Sometimes we have to balance self-improvement against practicality. The idea is to learn on the job, on the fly, and in real time with getting actual work done. Sure, it isn't ideal, but this is a reality for many.

then skipping over the process and cherry-picking what you think you need will only lead to you being stunted and short-sighted.
Uh huh. That is the only outcome? No exceptions? And you're 100% certain?

Yeah it's great if you're completely business minded and don't care about anything outside profit margins and "saving time".
That's not the only scenario in which it can be great. And anyway, that's the nature of many businesses out of raw necessity. Sometimes you gotta focus on the bottom line to stay afloat. I didn't write these rules and I didn't invent crapitalism.

But for the humans in the group that sort of attitude is detrimental psychologically.
People being focused on the company's bottom line is a matter of practicality and it does not equate to, for lack of a better term, a toxic work environment. As with most things in the world, there is a balance to be struck between being profit-driven + nimble while simultaneously fostering a top notch corporate culture with high retention rates and satisfaction among employees. This should be the goal and is not impossible to accomplish with sufficient planning & execution, plus a pinch of good luck and a healthy serving of hard work while you also "work smart" so to speak.

Nah. I'm not abashed to say I don't like it and I don't like what it is doing to both people and the way jobs are conducted, that it is a net negative, and that it will backfire spectacularly in short order.
Maybe you're getting old and crotchety, unaware of the grandpa style bellyaching you do now, being stubborn as a mule, blissfully apathetic about how you come across, pessimistic and yearning for some earlier time before the new fangle-dangled [insert tech trend].

It's short-termism that the public have the option not to play along with by choice; it's surprising to me just how quickly people forget about the consequences of other technology pushed on us in the past 20 years..
Yeah man, those fucking smartphones… so inconvenient. And higher vehicle safety standards? What the fuck is that shit? Oh no, did Facebook try something that flopped? How fucking dare they attempt to innovate or revolutionize a thing!? Touchscreens? Yuck. Cellular broadband? Gross. Bluetooth ubiquity? Pure garbage. Better USB standards? Stop pushing this tech on us! I yearn for ditto machines and dot matrix printers, Windows 3.1 / DOS, floppy disks, Leisure Suit Larry In The Land Of Lounge Lizards, and rampant bigotry. High-pixel density screens can fuck off, too. For that matter so can LED advancements, 3D printing, iPads, Drones, CRISPR gene editing, you get the picture.

like, you want people to be even more short-sighted and intellectually dim? Because that's what is going to happen.. and for what?
You have no faith in the younger generations. They're smarter, kinder and more savvy than we ever were. I'm optimistic about them, and I appreciate their retro obsession.

So we 'saved time' doing stupid shit for peanuts anyway?
Oh you haven't heard? Nobody gets out of this alive. You're taking life too seriously again.
 
Last edited:
This has went above my basic understanding of --- I do not understand it or know what it is capable of ---- natural reaction fear. Reading some of these things about it just recycling shite is good to read. Those articles not so much but no pretense "I do not understand it therefor fear it" is my .02 cents.
 
Oh you haven't heard? Nobody gets out of this alive. You're taking life too seriously again.
Just gotta enjoy the ride, learn some neat shit, do some neat shit, help people while teaching them and, importantly, learning from them

We'll get through it if we work together and the future will always be better as long as we never stop learning and teaching
 
it's tangential, maybe even just off topic, but i am reminded of one of my favourite bl quotes of all time:

roliepolie said:
I didn't spend the last 6 years of my life learning how to mix records, so that some 15 year old kid can come and take over the world with his fucking laptop and a pirated copy of Ableton...

from this thread: When Will Live Audiences Be Ready For Computer Mixing

:)

alasdair
 


I posted this a while back but am unsure if people saw it. Adam Curtis suggests that computers have actually been doing a LOT of the analyses of human behaviors long before AI became a buzz term.
 


I posted this a while back but am unsure if people saw it. Adam Curtis suggests that computers have actually been doing a LOT of the analyses of human behaviors long before AI became a buzz term.

Hypernormalization and Can't Get You Out of My Head are amazing - I also recently watched All Watched Over By Machines of Loving Grace by him as well.

He does a fantastic job of exploring these complicated and nuanced topics with style, humor, great music, and care. Thanks for sharing this!
 
Hypernormalization and Can't Get You Out of My Head are amazing - I also recently watched All Watched Over By Machines of Loving Grace by him as well.

He does a fantastic job of exploring these complicated and nuanced topics with style, humor, great music, and care. Thanks for sharing this!

The funny thing is that Adam Curtis has been making use of the hundreds of thousands of hours of BBC recordings that were previously just sat in an archive, often for decades. The way he's able to piecew togeth what is more a collage than a picture of history is something that deeply impresses me. He began in the 80s with his stuff on late-night slots but it seems like finally people recognize his unique take on events.

There are actually a lot of them and I've yet to be disappointed.

Are you a Jonathan Meades fan as well? All 35 episodes of his BBC one-offs and short series were never released on DVD BUT meadesshrone.com has them all. It's all done with a nod and a wink (the music usesd hadn't had copyright clearance).

 
The funny thing is that Adam Curtis has been making use of the hundreds of thousands of hours of BBC recordings that were previously just sat in an archive, often for decades. The way he's able to piecew togeth what is more a collage than a picture of history is something that deeply impresses me. He began in the 80s with his stuff on late-night slots but it seems like finally people recognize his unique take on events.

There are actually a lot of them and I've yet to be disappointed.

Are you a Jonathan Meades fan as well? All 35 episodes of his BBC one-offs and short series were never released on DVD BUT meadesshrone.com has them all. It's all done with a nod and a wink (the music usesd hadn't had copyright clearance).

I'm unfamiliar with Meades but I'll take a look!
 
I used GPT4 today to try and make some JSON data again. I asked it to generate some descriptions for the data set, unique descriptions. Five times I tried to explain, in different ways, why it was not providing unique descriptions - it kept putting the 'title' in the description and was obviously trying to pass that off as unique (technically it is obviously), despite every description being 99% identical.

It just can't hack it. If it were as powerful as people are claiming, it should have no trouble generating 100 unique descriptions for a data set like the one I was making, The context of the data set allowed for a massive variation to be had within the descriptions. It managed to generate unique 'titles' for the tasks, but it just could not do the descriptions no matter what I prompted. If it were so powerful, it could literally just shuffle words around in the descriptions.. like a plagiarist might do, or use different adjectives, etc.

Also, the image memes I made for the other threads (now deleted).. I got GPT to violate its own service policy. It told me I could not include political logos - I asked it to put an antifa badge on the cat - and the following iteration when I removed that request from my list, it then included it on the image lmao.
 
This was the last prompt attempt I gave it before just settling with what it churned out. I had argued with it 4 times before giving up. It followed this prompt, getting the general structure correct but it just refused to generate truly unique descriptions.. they were identical except for the embedded title within the description.
The description should be approximately 100 words, spread over two paragraphs. It should be unique amongst all the other task descriptions, it's content and structure suited to describing the task ("title"); it could be describing what needs to be accomplished by the designated employee, who they need to collaborate with, the benefits of achieving it in relation to the exposition, and so forth.
 
I gave it the JSON structure, along with a solid context for the whole data set i.e. an exposition event etc (not going to post details). It understood everything else, it generated unique titles for each item in the set which were relevant, it just could not grasp the description.
 
OK, so I decided to give another site a go (deepai.org). It did a much better job than GPT with the same prompting. But..

I was getting it to do batches of 5 descriptions. Working ok. 10? It gives me 9. No problem, "please output the next 5, starting at #39 in the array" (I provided it an array of the titles). It just started churning out titles of its own, but the descriptions were ok. 4 attempts at arguing with this one, and it still doesn't understand what it is doing wrong.

This is why it's inevitable that there's going to be some corporate or life critical disaster, due to people relying on these systems. Too easy for it to churn out data that is correct at face value only, and there's no way people are going to proof read its entire output.. I'm bored of doing that after 10 minutes lmao.
 
Top