Careercow Elon Reeve Musk - Tesla, SpaceX and Twitter owner + ex-paypal CEO. Manchild, sexual deviant, spergy autist with access to space travel

[888Flux]

Scolding a computer program like its a child. Beyond clown world.

What LLMs do is predict the next word (token) in a sequence based on a series of probabilities. It's very easy for people, even retards like myself, to understand how they work if you just put a little effort into actually looking it up.

3blue1brown has an excellent video on this subject. The math behind LLMs isn't all that complicated either. Chances are if you took College Algebra you can understand it.

People without computer literacy will anthropomorphize the output of LLMs because it resembles natural language. In reality it's a two-way search engine. We are far from general purpose "AI", and what modern LLMs are is a trillion dollar magic trick. I am not surprised at all that when you throw the entirety of the Internet (and more) at a computer model it is capable of producing said results.

It frustrates me to no end to see how overhyped and misunderstood this technology is. And just how much it's trying to be shoehorned into things where it's an ill fit. When you live in the funny money economy where nothing has any meaning anymore I guess that's how it goes. It's like everyone around you is constantly living in a manic episode.

 

[Hopkinsville Goblin]

What LLMs do is predict the next word (token) in a sequence based on a series of probabilities. It's very easy for people, even retards like myself, to understand how they work if you just put a little effort into actually looking it up.

3blue1brown has an excellent video on this subject. The math behind LLMs isn't all that complicated either. Chances are if you took College Algebra you can understand it.

People without computer literacy will anthropomorphize the output of LLMs because it resembles natural language. In reality it's a two-way search engine. We are far from general purpose "AI", and what modern LLMs are is a trillion dollar magic trick. I am not surprised at all that when you throw the entirety of the Internet (and more) at a computer model it is capable of producing said results.

It frustrates me to no end to see how overhyped and misunderstood this technology is. And just how much it's trying to be shoehorned into things where it's an ill fit. When you live in the funny money economy where nothing has any meaning anymore I guess that's how it goes. It's like everyone around you is constantly living in a manic episode.

Sad thing is that kids now use AI to do as much of their work as is allowed. I've argued with some online who feel this is fine. They don't have to learn anything because they can just ask the AI. So they lack the computer literacy and math skills to understand the AI's workings and they lack the critical thinking skills to see when it's totally wrong.

This is a similar problem to when calculators were introduced. Lots and lots of students never fully understand the four basic arithmetic functions of a simple calculator. So they can plug in 3 + 8 and find 11, but they have no idea how the plus sign relates those two numbers and combines them to 11. I've seen high school students at this level. 27/3 can easily be calculated as 9, but there is no understanding that that means the 27 got split into three pieces each worth 9.

So by the time they're in high school the teacher's expected to get them to grade level. So at 11th grade, that should be about precalculus where I was, or calculus 1 for the smarter ones. Let's say they're starting prealgebra, where I was in 6th grade. The problem looks like this:

Solve for x: 2x + 5 = 15

So you can hold their hand and get them to subtract 5 from both sides. That requires the calculator, of course, for 5-5 and 15-5. Then they have 2x = 10. At this point the student considers the problem done and is already typing in the next problem:

Solve for y: 3y + 9 = 21

So at the point the student is already writing 12 as the answer because they typed in 3 + 9. And if you tell them the first problem still isn't complete, they say "but I did it."

Now imagine you've got 30 kids like this. And they're getting driver's licenses.

I've messed around with Brave AI and I've seen it flub up things like converting kilograms to pounds. It can't get significant figures right either - like let's say I ask it for the approximate weight of apple in ounces. Maybe it would say approximately 6 ounces. Then ask it to convert to metric units. Suddenly the answer is calculated down to the hundredth of a gram.

So anyways I think our schools should go back to pencil and paper, lock up the smartphones, and instead of calculators go back to the fucking abacus. This is not as obsolete an idea as it sounds. I've seen videos of Asian schools were they do that and eventually the kids don't even need the abacus because they can multiply two 8 digit numbers in their head, just operating an imaginary abacus in the air with their fingers. For more advanced math, you can still use the abacus, or learn to use a slide rule. This forces a student to actually understand math. They can have a TI83+ when they get a job.

For geometry, you have to draw the problem yourself with a pencil.

If you're working from a book then copy it onto paper.
If you're working from a computer, copy it onto paper.

This is like pulling teeth but I've tutored F students in geometry and got them up to a B in a matter of days if I can get them to draw. They have to draw the shapes but that seems too difficult for them. Here is an example of the type of drawing I'd want them to copy onto paper. They are always like "I caaannnn'ttt".




I'll leave it as an exercise for the reader to solve these problems to completion. You should be able to do them all in your head if you've got a high school diploma or GED. If you need AI, go back to school. Don't get me started on reading and science.

EDIT: I'm sure the Matrix squid robots are watching me, but I just got recommended this video. I recommend you all watch it. This is horrible. When the teacher wanted those two guys to spell "philosophy" they at least did that phonetically, and I've heard of worse. I urge any American with even the most passing interest in the future of education to listen to the podcast Sold a Story.

https://www.youtube.com/watch?v=HGd7Mj7k97Y

 

[donksmanSupreme42]

For geometry, you have to draw the problem yourself with a pencil.

i always found paper and pencil to be more comfortable than typing on a computer, i write like 20 times faster than i type, and i think like 10 times faster than i write

the only things i ever used a calculator for were trig functions

 

[Hopkinsville Goblin]

i always found paper and pencil to be more comfortable than typing on a computer, i write like 20 times faster than i type, and i think like 10 times faster than i write

the only things i ever used a calculator for were trig functions

I had to look those up on a chart, as well as things like p values in t and chi squared tables in statistics. My textbooks came with some of those printed in the inside front and back covers, and there was an index. A calculator was certainly helpful, but I could grind through it easily enough on paper.

It's trivial to make such tables now with Excel if you need to. But a long time ago, the computers who laboriously made all those tables were women. That was their job title, computer. Those women understood the math they were doing. Grace Hopper must be spinning in her grave.

 

[AnOminous]

It frustrates me to no end to see how overhyped and misunderstood this technology is. And just how much it's trying to be shoehorned into things where it's an ill fit.

The problem with AI isn't what anti-AI seethers complain about, it's what stupid people think it is and what they do with it. Relying on it to make complex decisions from a human point of view is outright insane. You can get it to say what you want it to say even inadvertently and not knowing you're doing it, meanwhile it's woefully biased toward the currently "accepted wisdom" even when the "wisdom" is dumb as dogshit.

Never mind the fact it's gimped not to say certain things regardless of their truth or falsity.

Musk is aware of this, but the dumb shit he constantly says about it shows he honestly has no more clue about what it's actually doing than even the average Kiwi with no particularly special experience.

 

[888Flux]

The problem with AI isn't what anti-AI seethers complain about, it's what stupid people think it is and what they do with it. Relying on it to make complex decisions from a human point of view is outright insane. You can get it to say what you want it to say even inadvertently and not knowing you're doing it, meanwhile it's woefully biased toward the currently "accepted wisdom" even when the "wisdom" is dumb as dogshit.

The problem with AI you refer to is that it's not AI. It's not doing any thinking besides putting text the model was trained on together in a sequence. LLMs aren't capable of cognition. A text model has no bias because it works in the text domain. To the model, a word is just a vector representation of a token with a numerical value.

The model is a text predictor. It predicts text. It's entirely a black box and the model has no understanding of what that text actually means. The reason why LLMs can do anything that resembles "problem solving" is because of the sheer volume of data the model was trained on. If you watch the 3blue1brown video linked in the previous post it will explain how it works perfectly.

Actual general purpose artificial intelliegence that can work in a domain higher than text is probably never going to happen, and I think part of that is due to the slop-ouroboros and digital Kessler effect spawned by the mass proliferation of LLMs to countries like India and people like Elon Musk.

 

[AnOminous]

The problem with AI you refer to is that it's not AI. It's not doing any thinking besides putting text the model was trained on together in a sequence.

I agree but that's what we're calling it even though it isn't. (The fact some LLMs seem to be able to pass some variants of the Turing test doesn't really show intelligence but that the Turing test is not really a test of intelligence.)

Actual general purpose artificial intelliegence that can work in a domain higher than text is probably never going to happen, and I think part of that is due to the slop-ouroboros and digital Kessler effect spawned by the mass proliferation of LLMs to countries like India and people like Elon Musk.

I think it will but in nowhere near the timeframe messianic AI worshippers (and hysterical doom-mongers like Yuddo the Clown) predict, and it won't be an LLM.

 

[donksmanSupreme42]

I had to look those up on a chart, as well as things like p values in t and chi squared tables in statistics. My textbooks came with some of those printed in the inside front and back covers, and there was an index. A calculator was certainly helpful, but I could grind through it easily enough on paper.

yeah, i did memorize the commonly-used "trivial" values tho

 

[Fox Girl Enjoyer]

It's a nice start. They seem to be scaling up towards where many Chinese companies are already. If they can deliver extreme reliability they might be able to get a niche where they are as ridiculously expensive as Elon's shitware but much less likely to explode, and much more expensive but slightly less likely to explode than the Chinese options.

Didn't realize China was making their own rockets. But I could see how the East just starts blowing past us in a lot of these high tech areas (well more so than normal) due to competition from China. Say what you will about the Chinese but those 5 points matter and trying to fix the problem with Indians is going to make it worse. My apologies to any Wallstreet investors who are banking on THIS TIME being the time India becomes the next big thing.

Actual general purpose artificial intelliegence that can work in a domain higher than text is probably never going to happen, and I think part of that is due to the slop-ouroboros and digital Kessler effect spawned by the mass proliferation of LLMs to countries like India and people like Elon Musk.

Current AI is more for Wallstreet than it is for anyone else (the only real skill Elon has is getting their money). I think you're right and if AI is to ever become a thing it wont be through the LLMs. Its certainly not coming from fucking America thats for sure. We lost our edge in the 2000s and now with the rush to hire Poojeets we have ensured we won't have a generation of computer guys that could make the next big breakthrough. We'll see the trickle down affects before long. Allowing Indias into tech was one of the worst decisions the US could make

 

[Dan Abrams]

Apparently Elon Musk does not in fact use a computer.

 

[AnOminous]

Didn't realize China was making their own rockets. But I could see how the East just starts blowing past us in a lot of these high tech areas (well more so than normal) due to competition from China.

China and India are both low trust societies but China at least has some traditions (despite the Communist attempt to destroy them and replace them with nihilism) that they could revert to and become a high trust society. India is just always going to be a bunch of fucking jeets.

Low trust societies are worthless when it comes to reliability and all they can ever really offer is being cheap as shit by using near slave labor. Try buying raw materials from chinks. You'll specify some specific grade of steel and they'll deliver some chinesium pig iron chink shit every time.

 

[Deep Nozzle Throat]

This is going to veer into off-topic I'm afraid, but watching a crane fall over during a lift operation at Massey's in Texas has nudged me to cave and start talking about just how bad Musk's lolcowdom has screwed over SpaceX, which I have some insight into.

Musk was pretty hands-off at SpaceX through most of the Falcon 9/Heavy and Dragon dev days, because his real pet was what was originally called the Big Fucking Rocket (BFR). The thing that most people outside of the industry don't really get is that the former's success was almost entirely due to Musk picking rising stars and putting them in the right roles. You may have heard of Tom Mueller, who led the charge on the Merlin family of rocket engines, and Hans Koenigsmann, who was the lead on Falcon 9 reliability and mission assurance. Somewhere between 2016 and 2020, where BFR would become a carbon fiber monstrosity called the Interplanetary Transport System, both of these people and some others were quietly forced out. Mueller was put into a backwater role on Raptor engine production and Koenigsmann came to metaphorical blows over the trajectory of BFR/MCT/ITS/BFR/Starship as a dev program. I utterly despise NASASpaceflight, but this article from a few years back details how Musk slowly centralized direct control over the program.

Falcon Heavy is a critical launch vehicle for many of their defense flights. Though it was originally heralded as the transformative rocket for the space industry before a delayed debut in 2018, it pretty clearly never met those ambitions (having flown 10 times since), and Musk himself tried to cancel it before being convinced otherwise. Axing Dragon amid the catfight is absolutely the kind of thing that's been discussed behind closed doors.

Those of you thinking Starship as the Cybertruck of SpaceX are more right on the money than you'd think. The overall design is somewhat questionable - I've seen some evidence that stainless steel in particular doesn't like high-vibrational environments, and this was a concern when Centaur was investigated for use and reuse as a space tug during the early Shuttle program - and a vicious cycle exists between the physical airframe being too heavy and the engines being pushed to greater and greater performance limits. Starship Block 2, which made Flights 7, 8, and 9, and then hardware for 10 blew up during testing, is really struggling. Unclear as to what the root cause is, but the inside view is that it's a combination of quality control and Musk descending from on high to push for the next launch no matter how unprepared the system is. There's many Band-Aid fixes that go into each vehicle. Quality control is somewhere between abysmal and nonexistent, and there's a tremendous reliance on individual heroics to deal with issues rather than any kind of process control.

The worst part about this is that aside from the direct responsibility SpaceX has for the HLS contract, there's been a serious push to gut NASA's existing launch capability and replace it with Starship. Obviously Musk pissed off Trump at the White House, and Jared Isaacman got withdrawn because he was too close to Musk. Congress is pissed about the whole thing and the string of failures plus the Musk/Trump divorce has really shut the door on a regulatory capture gambit that almost got pulled off. Internally a shitshow is brewing at NASA.

Musk's autistic obsession with Mars is starting to have consequences with Falcon and Dragon too. Dragon has been flagged for a significant number of waivers (more than you'd be comfortable to know) and the extant Falcon vehicles are starting to show their age. Axiom-4 almost launched with a cracked LOX line that would have probably blown up in flight. There's some concern inside and out that as engineers get pulled from Falcon and Dragon to go work on Starship the remaining crew loses a lot of institutional knowledge about the vehicles. I have personally witnessed SpaceX employee burnout and it's never a pretty sight. I'm genuinely concerned that within a year there will be a loss of mission (LOM) based on creeping trends.

All that to say that Musk the spergy fuck has singlehandedly taken a revolutionary company with incredible capabilities and gutted it just to fund his tardy Mars fixation, and a shrinking but virulent army of jannies keep running defense for him. The blind headlong rush is going to get people hurt or killed.

 

[Hollywood Hulk Hogan]

@Deep Nozzle Throat your username definitely checks out. That's interesting

 

[Deep Nozzle Throat]

@Hollywood Hulk Hogan True to the name, I'm just a very worn-out individual who's been privy to a lot of insight on spaceflight in the US.

What I'm most worried about overall is that Musk has led to a lot of cargo-culting in commercial space companies. A lot of them play fast, loose, and dangerous, and they don't have the reality distortion field to convince investors to stay on so they end up bankrupting themselves during development and testing. It's at times enraging to see traditional companies with robust dev and safety guidelines to get thrown under the bus by both the public and other industry workers because they avoid corner-cutting and lean hard into safety. The fact that Starliner is remembered as "stranding two astronauts" is a prime example of that. Not to exonerate Boeing as they too have problems, but their space sector is nothing at all like their aviation sector and they overall run a safety-conscious and otherwise tight ship with a very high degree of transparency. That's bit them in the face of SpaceX.

The more frustrating thing (again, drifting into off-topic) is that the army of jannies has almost singlehandedly wiped out any kind of productive conversation regarding spaceflight and has actively chased industry experts out of most of the conversation spaces here on the 'net over phonebooking and actually bad things. The community of space enthusiasts almost deserves its own thread, but it's niche enough that I don't think it'd gain much traction.

Fortunately I think the darkest days are behind us. There were definitely quite a few journoscum (and I don't say that lightly) that were ready to push "SpaceX's Starship poised to supplant SLS" articles around the time Flight 7 exploded, but God willing those articles continue to languish on hard drives and in drafts. Communities have been falling apart as skepticism has creeped in and it's no longer wholly verboten to say that the program is doing poorly and that Musk is a sycophant. But the damage has certainly been done. It seems unlikely the OMB budget is going to get approved and that basically was the last real chance to terminate SLS/Orion as EUS work is too mature to get yanked after about next year. Artemis III (the intended lunar landing) is probably getting rebaselined though, those conversations have started back up.

I'll be watching AX-4 tonight with white knuckles though. Really hope that goes OK, that particular Dragon has been very troublesome. The original proposed fix was to just install a nitrogen purge (this is not standard industry practice but exceptionally common on Starship, with both stages featuring purges to try and get ahead of leaks that keep forming in flight).

 

[James done did it]

Even if starship fucking worked the entire mission architecture for even a moon mission is moronic, let alone a mars mission. How many successful launches is it going to take to get one starship to Mars? Even if it worked and resources weren't a concern, Mars is basically a pointless endeavor. It offers literally nothing the moon does not offer and comes with massive additional challenges. The fact that it has higher gravity and more of an atmosphere is not a benefit, neither is enough to prevent longterm and acute health issues for humans. In fact it's an impediment because it increases the challenge of landing and the cost of launching. The atmosphere, increased distance from the sun and dust storms also massively hinder solar power generation relative to what's available on the moon.

The moon is worth colonizing because it acted as a massive collector of asteroids and is a place where we can stripmine without worrying about the environment. The earth isn't running out of minerals but it's running out of minerals that we can exploit on a scale that keeps the global economy humming without eventually crashing ecological systems. The asteroid belt is far too diffuse for practical mining and much farther away. Mars is in a similar situation but the extra distance, gravity and atmosphere all mean it's just never going to be economical while the moon remains the lower-hanging-fruit.

 

[Something Awful]

I have to ask, has anyone experienced Twitter's "For you" recommendations feed getting significantly worse recently?

(Yes, yes, Twitter has always been and will always be shit, but for a while I was seeing tweets that I found somewhat funny or informative at least reasonably often)

Recently it's all just pure brainrot retardation and extremely low-IQ clickbait/reply bait. Like, it's all this:




Is it just me, or did they actually intentionally make the algorithm even worse and dumber, bro?

 

[World in Scarlet]

Is it just me, or did they actually intentionally make the algorithm even worse and dumber?

My "For You" occasionally can't tell the difference between slop and quality coverage of the same topic, but 99% it's relevant to my interests.
I mostly use the "Follow" tab, maybe that prevents me from seeing enough clickbait to poison the algorithm.
I also liberally block accounts that post garbage.

---

link
AI delusions again.

 

[Deep Nozzle Throat]

Even if starship fucking worked the entire mission architecture for even a moon mission is moronic, let alone a mars mission. How many successful launches is it going to take to get one starship to Mars? Even if it worked and resources weren't a concern, Mars is basically a pointless endeavor. It offers literally nothing the moon does not offer and comes with massive additional challenges. The fact that it has higher gravity and more of an atmosphere is not a benefit, neither is enough to prevent longterm and acute health issues for humans. In fact it's an impediment because it increases the challenge of landing and the cost of launching. The atmosphere, increased distance from the sun and dust storms also massively hinder solar power generation relative to what's available on the moon.

The moon is worth colonizing because it acted as a massive collector of asteroids and is a place where we can stripmine without worrying about the environment. The earth isn't running out of minerals but it's running out of minerals that we can exploit on a scale that keeps the global economy humming without eventually crashing ecological systems. The asteroid belt is far too diffuse for practical mining and much farther away. Mars is in a similar situation but the extra distance, gravity and atmosphere all mean it's just never going to be economical while the moon remains the lower-hanging-fruit.

The situation with refueling tankers (if anyone uses the word "refilling" it's a shibboleth for Muskcucks) is that it's going to take a lot of launches just to complete HLS, let alone go to Mars. Boil-off is a serious problem on vehicles of that size. Back when the Saturn V was being investigated for Mars missions, the best boil-off rate that could be accomplished on paper was 1 metric ton of prop per day. So the end result is you're stuck with an Operation Black Buck tier refueling process, with tankers being spammed in LEO and then sending the HLS vehicle to an elliptical high Earth orbit to meet a depot that, likewise, has been loaded with prop from other tankers. It's in the high teens if not more at this point.

That's all compounded by the fact that Starship is essentially the upper stage of a two-stage vehicle, and that means if it wants to go anywhere it has to haul around a lot of dead weight. Every HLS proposal has a refueling problem, but it's especially bad with Starship because it's so damn big that it demands a ridiculous long tail of prop launches. (The rocket equation is logarithmic).

And to the topic of the thread this is entirely due to Musk's own obsession with this being the "correct" solution for the problem. An all-chemical Mars mission hasn't been seriously investigated or proposed (excluding Mars Direct) since about the 1980s. The Design Reference Mission for NASA uses nuclear thermal rockets, and the current vibes at the agency are to use solar electric or nuclear electric propulsion, both of which are leagues more efficient and fault tolerant than all-chemical (somewhere between twice and ten times the performance of methane/LOX).

 

[thejackal2]

I'm totally not on drugs see!

I have to ask, has anyone experienced Twitter's "For you" recommendations feed getting significantly worse recently?

(Yes, yes, Twitter has always been and will always be shit, but for a while I was seeing tweets that I found somewhat funny or informative at least reasonably often)

Recently it's all just pure brainrot retardation and extremely low-IQ clickbait/reply bait. Like, it's all this:

View attachment 7553543 View attachment 7553544 View attachment 7553545
View attachment 7553546 View attachment 7553549 View attachment 7553550 View attachment 7553551 View attachment 7553547

Is it just me, or did they actually intentionally make the algorithm even worse and dumber, bro?

Out of fucking nowhere I'm now getting "25yearsago" or 125yearsago" junk pushed into my feed. No I really don't need more gimmick accounts pushed in what I am trying to use for near real time news and reaction. I hate what he's done to Twitter it always was flawed but Elon just whored it out to slop and scams and fake news.

My "For You" occasionally can't tell the difference between slop and quality coverage of the same topic, but 99% it's relevant to my interests.
I mostly use the "Follow" tab, maybe that prevents me from seeing enough clickbait to poison the algorithm.
I also liberally block accounts that post garbage.

---

View attachment 7553562
link
AI delusions again.

His brain must be completely fried from the drugs. He should watch some growling sidewinder on youtube or some shit. How is a pure vision (so visible spectrum) missile going to hone in on a target that's over 30 miles away? It ain't. Which is why it uses a full sensor package. This retard applies the same logic to his self driving cars and it's not working. You need a full sensor package. Visible spectrum is prone to all sorts of issues, including dust, glare, clouds, simple decoys etc.

 

[Drip Torch]

Of course stealth doesn't mean invisible. But autists think they know everything.