Notice: You have been identified as a bot, so no internal UID will be assigned to you. If you are a real person messing with your useragent, you should change it back to something normal.
Anonymous A (OP) replied with this 1 month ago, 3 minutes later, 10 minutes after the original post[^][v]#1,436,924
@previous (B)
I haven't seen it demonstrated, but it is surely within the reach of deep learning and mechatronics today if someone did try to build it.
Mr. Black Boi double-posted this 1 month ago, 1 minute later, 3 hours after the original post[^][v]#1,436,942
To be fair, would they cover it in the media if a human solved a math problem? People only care about a discovery if they can immediately identify how the discovery is useful to them, if they can’t, then they don’t care.
Mr. Black Boi replied with this 1 month ago, 1 minute later, 3 hours after the original post[^][v]#1,436,954
@previous (A)
If you think an autocomplete algorithm is smarter than you, you’re probably right. You can’t even spell the n word correctly. You don’t even have the IQ required to be an effective racist.
Anonymous F joined in and replied with this 1 month ago, 8 hours later, 12 hours after the original post[^][v]#1,437,009
Its hard for me to follow AI. I get the benefits, I use it, but the hype machine is accelerating because there is a bottleneck of funded companies whose investors want to exit. It’s 1999 again.
Only this time the change is bigger. If there’s a huge societal shift there is going to be some degree of chaos. If millions are dispossessed, why would they continue to give a shit about society? Follow the rules, get burned, go extreme.
Anonymous A (OP) replied with this 1 month ago, 3 hours later, 15 hours after the original post[^][v]#1,437,030
@1,436,954 (Mr. Black Boi)
I don't know anyone, of any race, solving multiple old math problems.
You can't "autocomplete" a sentence that has never been written. That should be obvious and common sense. Just like it should be obvious that you can't reverse a hash function that has more inputs than outputs, or that it should be obvious that logic in a debate is different than computational logic.
Spamposting compsci trivia slop doesn't mean you understand any of it.
Mr. Black Boi joined in and replied with this 1 month ago, 10 minutes later, 16 hours after the original post[^][v]#1,437,033
@1,437,030 (A)
You can autocomplete a sentence that’s never been written, you don’t even need a sophisticated algorithm to do that.
Imagine taking three sentences:
"There is a red house."
"Fire trucks are red."
"Mathematicians are smart people."
Now imagine you have an algorithm that takes pairs of words that appear next to each other in the data.
(There, is)
(Is, a)
(A, red)
(Red, house)
And so on for the other two sentences.
Now imagine I write this:
"Are"
Well now imagine the autocomplete algorithm keeps randomly adding a word to the end based on pairs that appeared in the data:
"Are red"
"Are red house"
Was "are red house" a sentence that was in the data that the algorithm was trained on? No…
Now imagine a much more sophisticated algorithm that uses a neural network. Obviously that claim that an autocomplete algorithm can’t generate a new sentence that’s never been seen before is completely false.
Anonymous H joined in and replied with this 1 month ago, 1 minute later, 16 hours after the original post[^][v]#1,437,034
Ugh who cares. That is boring nerd shit. People will care when AI does something actually cool, like blowing up the moon or making women's boobs bigger or something.
Mr. Black Boi replied with this 1 month ago, 4 minutes later, 16 hours after the original post[^][v]#1,437,035
Then if you look up how many words there are in the English language, which actually isn’t a thing everybody agrees on because different words have multiple words and some words have multiple spellings so it’s not obvious how those should be counted so different people come back with different. Regardless, if you use the permutation formula, not every combination of words will be grammatically correct, but nevertheless, if you increase the length of a sequence of words, the number of possible permutations given all the words in the English language will increase rapidly the longer that sequence is. To the point, you’ll very quickly get numbers that a calculator or even a 64 bit computer can’t represent. Then if you factor in, LLMs don’t actually understand words, they’re trained on tokens and each token is a vector in a high dimensional space which roughly correlates with meaning, and there are more tokens than there are words, and LLMs are generating millions of tokens for cents in some cases, the probability that they’ll be generating sentences that have never been written by a human is essentially 100%.
Mr. Black Boi double-posted this 1 month ago, 1 minute later, 16 hours after the original post[^][v]#1,437,036
@1,437,034 (H)
Ngl, that’s one of the things I don’t like about the current direction of AI. There are other paths we could take technology that would be way cooler than slop machines that spit out text that convince mentally unstable people they’re talking to God, like space travel. Space travel would be cool af. I want Star Wars IRL - maybe minus the war part
Mr. Black Boi triple-posted this 1 month ago, 2 minutes later, 16 hours after the original post[^][v]#1,437,037
Usually they use the cosine similarity because it’s more computationally efficient, but the Euclidean distance formula is just the square root of the dot product of a vector with itself. It just so happens that space is 3 dimensional, but you can use any arbitrary number of dimensions and it still works, so you can use Euclidean distance to measure the similarity of two words that have been vectorized by an LLM.
Anonymous A (OP) replied with this 1 month ago, 2 minutes later, 16 hours after the original post[^][v]#1,437,039
@1,437,033 (Mr. Black Boi)
It's coming up with solutions and proofs for math puzzles no human could solve.
If that's autocomplete, then autocomplete is a powerful tool that can replace human reasoning anywhere. Is there any real empirical evidence that the brain is doing something besides predicting the next words or images?
Giving it a dismissive label like "autocomplete" doesn't identify a single testable difference between the two.
Mr. Black Boi double-posted this 1 month ago, 3 minutes later, 16 hours after the original post[^][v]#1,437,041
Don’t get me wrong, computers can do math, it’s cool. Computers have been able to do math since they were invented. And even before LLMs, humans wrote very sophisticated algorithms that could solve very complex math problems with a high degree of accuracy. If an LLM solves a math problem better than a human does, that doesn’t convince me that it’s more like us than a computer… especially given the fact that it literally is a computer. AI is just a scam by the linear algebra industry.
Anonymous I joined in and replied with this 1 month ago, 1 minute later, 16 hours after the original post[^][v]#1,437,043
@1,437,041 (Mr. Black Boi) > AI is just a scam by the linear algebra industry.
Global AI investment reached $581 billion in 2025, surpassing the prior record of $360 billion set in 2021. U.S. private AI investment climbed to $285.9 billion in 2025, up 162% from $109.1 billion in 2024. The U.S.-to-China investment multiple widened to 23x in 2025, compared to 11.7x in
Soon to hit over a Trillion of $$$$$$$$$$$$$$
I am looking forward to what it can do with cancer research.
Mr. Black Boi double-posted this 1 month ago, 2 minutes later, 16 hours after the original post[^][v]#1,437,045
@1,437,043 (I)
The thing about the valuation of a company, is the way investing works, is someone puts money into a company to buy a slice of the company with the hopes that somebody else will come along after them and pay more for it than they did. But there’s bidding going on, in order to buy a slice of the company you have to compete with everybody who wants to buy that same slice, which drives up the price people are willing to pay.
Notice how none of that has anything to do with whether or not the company is selling their product for more than it costs to make the product.
Mr. Black Boi triple-posted this 1 month ago, 3 minutes later, 16 hours after the original post[^][v]#1,437,046
Really, the valuation of a company doesn’t actually mean that much. It’s based on speculation. Part of the reason why AI companies that haven’t gone public have such high valuations is because private companies don’t have to publicly report detailed financial information to the public. So for example, SpaceX has a high valuation, and many people thought SpaceX was the future, then when they filed for their IPO recently, we all found out they’re essentially burning billions of dollars. When OpenAI IPOs, I’m just going to say, they’re probably worse than SpaceX. Does that mean people won’t invest in the short term and the stock won’t go up in the short term? No, people probably will invest in both OpenAI and SpaceX on the public stock exchange, because people are irrational. However, the easiest way to lose money in the stock market is to do what everybody else is doing, because most people lose money. Most people won’t get out in time.
Mr. Black Boi quadruple-posted this 1 month ago, 7 minutes later, 16 hours after the original post[^][v]#1,437,047
I’d argue the higher the valuation is already before you buy, the more suspicious you should be that someone will pay you even more for your shares later. Investors have a lot of money, but not an infinite amount. When you start talking about trillion dollar valuations for some of the most unprofitable business models that anyone has ever thought of, that’s going to be a train wreck at some point.
Mr. Black Boi double-posted this 1 month ago, 5 minutes later, 1 day after the original post[^][v]#1,437,133
If you want to convince me that sand is intelligent, what’s sand made from? Silicon dioxide. What’s a computer chip made from? Silicon dioxide. Idk. Transistors are just on/off switches. Neural networks are just software. An LLM is just software.
Mr. Black Boi joined in and replied with this 1 month ago, 2 hours later, 1 day after the original post[^][v]#1,437,156
@previous (A)
My point is that inanimate objects aren’t sentient. Whether you make it from graphite or silicon or whatever, a computer is an inanimate object.
Mr. Black Boi double-posted this 1 month ago, 7 minutes later, 1 day after the original post[^][v]#1,437,157
LLMs also aren’t remotely the same as human brain. Only a relatively small part of your brain is dedicated to language. LLMs also can’t learn while they’re inferencing at the same time. Humans learn continuously while they’re performing actions. Even on the level of a perceptron, a perceptron is based on a flawed understanding of how a biological neuron works, they’ve found that an individual biological neuron is far more capable than an individual perception in a neural network and individual biological neurons can perform tasks that are impossible for an individual perceptron to perform. Biological neurons also can perceive time on an individual level and understand signals recurring across different time intervals, but perceptrons can’t, they’re basically stateless.