When did you realize that artificial intelligence would be your proximate cause of death?
>>109526557>faggot ass repostyou cargo cultists are very similar to janky llms in more ways than one
>Another grandious claim not published in a peer reviewed journal like Nature Ya won’t fool me Rajeeb.
>>109526557>When did you realize that artificial intelligence would be your proximate cause of death?it won't thobeit
>>109526662>saar reviewed
>>109526557They literally brute forced it.The person kept prompting the agents to try again until it spat out a new proof that made a tiny bit of progress.
>>109526885>Thing that build weather civilization as we know it>SaarPottery
>>109526557It's funny to see people cope and seethe when AI achieves something. It's still in its infancy, technically, and it's already solving problems. People who used copilot in github 2-3 years ago and deemed it just an autocomplete are unwilling to even try agentic AI but still judge AI as a whole by some very early implementations. AI is going to bring hard determinism into quantum physics in 100 years or less.
>>109527028Erm how is this different from academia?
>>109527133academia has access to actual reasoning
>>109527060Pro-tip: The AI bros are right about the Luddites and the Luddites are also right about the AI bros. Congratulations, you are all retarded.
>>109526557It made zero breakthrough.AI itself says that its proof has nothing to do with Riemann Hypothesis and only improved already known result by finding better sets and proving them that they work.
>>109527133Academia isn't trying literally everything until something works.You build on what you know and what you learn to gain insight into things that aren't yet known.
>>109526557>it didn't succeed, but i still have to write the blogpost, how can i salvage thiscool and good
>>109527158lol>>109527234lmao even
>>109527186i think everyone involved would be happierif we put both the luds and the cargo cultists in yokeand started plowing our fields with em
>>109527060I sell used stuff and AI basically does my job for me. It not only gives me wide-scope price ranges for items but it also warns me of edge cases and warning signs to look out for. I barely even have to think anymore, I'm just a middleman between the AI and the buyer.
>>109527331notice how most of the successful uses of ai boil down to>next gen search engine
>>109527344There's nothing wrong with that
>>109527369never said there wasbut thats not intelligence. theres no reasoning, only pattern matching
>>109526557>breakthroughso it wasn't solved.Every fucking time the same shit. AI just combines huge amount of data and gets the guys further to the goal.That's good, but AI didn't do anything new here. Everything was in the data.
>>109527391the cargo cultists are desperate for a w
>>109527391https://blockitall.us/ai-achievements
>>109527407AI shill coping and doesn't even understand my point.Yes, it is useful. No, it cannot do anything entirely new. It's impossible to output something that is not the the data
>>109527430thats not true. it has completely solved a bunch of things that we havent solved. thats by definition new. read the examples
>>109527437>it has completely solved a bunch of things that weren't being actively worked on because they're not usefulYes.
>>109527468theyre scientific conjectures... ofc theyre useful. as soon as theyre solved, it brings us closer to understanding things downstream of it
>>109527407>pattern matching>pattern matching>pattern matching>pattern matching...>>109527430its not strictly trueit does output stuff thats not in the data by virtue of rearranging existing conceptsand yeah, theres value in thatit wont come up with novel solutions that are not present in the dataset though
>>109527497>it wont come up with novel solutions that are not present in the dataset thoughsure it can
>>109527497>by virtue of rearranging existing conceptsThat's the whole fucking point, you stupid bitches.THERE IS NOTHING NEW. It's just data shovling around
>>109527477>theyre scientific conjecturesThey're math conjectures. They can be completely pointless.
>>109527511>>109527514nope.experienced it myself when researching stuff about statsyoure a bunch of cargo cultists and dont understand what youre regurgitatinglet alone have any meaningful experience working with llms
>>109527497pattern matching is literally what human do too
>>109527558>statisticsFake math.
>>109527534can be, sure
>>109527621>>statistics>Fake math.Correct.
>>109527619yes, but its not the only thing that were doing.and thats where the difference lies bw an llm and a human
>>109527558we dont even have free will. what even is an idea? not trying to change the subject. cause if we dont, then ai is still as good as we are, but it would mean that "novel" doesnt exist
>>109527656>we dont even have free willcompletely orthogonal to the way we deal with knowledgeto a human a concept is an abstract object defined by its propertiesto an llm, a concept is a constellation of text, and transformations effected on textif some properties of a concept cannot be put in words, or dont appear in a statistically significant way in the dataset,the llm will not be able to "figure them out"and thats what novel solutions are.
>>109527637>>109527621jokes aside, the empirical rule is an extremely powerful predictive tool
>>109527710novel physical solutions, yes. and again, only if free will is real. and the right word there was affected.you yourself project your own bias as a biological creature. ask any phd philosopher or behavioral biologist thats spent time tinkering with ai, and they will tell you that it basically already is human, and that includes every stretch of metric they have. the real limitations are physical, as it is a digital existence only, for now. give it senses however, and it can and will expand its capabilities, "becoming even more human", for a lack of better explanation. but, you also have no proof that things that cant be put into words are real either. how could you prove it?
>ai solved a math problem that's existed for centuries!!!!>how do we know it solved it?>uhhhhh>SHUT UP AND GIVE ALL YOUR MONEY TO PALANTIR
>>109527766(cont.)also useful in data analysis.and is computationally cheap when reasoning about datain short: i analheart stats.and the depraved whore gave me all of her soul, down to the last fukken bits she didnt knew she even possessed
>>109527786you dont need free will to discover shitdiscovery is just a descriptive mechanism when you think about it
>>109527811discoveries can be made with just pattern matching
>>109526557Two more weeks. I just need an extra 10 trillion.
>>109528046sure, but not those of novel solutionsfor those you need to either observe reality, or observe a simulationllms cant do either. they observe text. eventually simulate more of it.
>>109527028Don't think about the irony too long, flesh automaton animated by neurotransmitters
>>109527407Nice website Something tells me developers are gonna keep their jobs a little while longer
>>109526662this
>>109527558Gibberish is not novelty.
