>>539244735AI does that in 1 second.
>>539244735>one sentence>no references
>>539244735is this referenced or peer reviewed? seems like somones bullshit opinion.
>>539245072You can't know when you're being lied to unless you do the research. AI just lies to you faster than humans can. There is no replacement for your own intellect.
>>539248514You can be lied to through the research papers as well.
>>539248887That's why OP suggests that reading 6 papers might result in 2 references. That's kinda the point. The robot doesn't even know it lies, it just has guardrails with no ability to discern what papers are legit and what disqualifies them.
>>539249099Actually AI has "prior knowledge" so to speak to be able to determine is a research paper is correct or fits with current scientific doctrine.
>>539244735AI can do that in 2 minutes.
>>539248514My AI agent shows its references, so I can just click and read the peer-reviewed studies it has used as source material.
>>539248887Both humans and LLMs can make mistakes. But the problem is that LLMs are just guessing the entire time. They are very good at guessing, but that's all they do. They are incapable of analytical thought. That's how you get "the letter M appears 3 times in the word tomato". If you entrust a stupid machine to do your research then you'll get exactly what you deserve.
>>539249320That's a really old hallucination I believe that has been fixed and they really do calculate now.
>>539249248Which is why it lies... AI works like females do - seeking consensus over truth. AI has no concept of truth at all, it can only check if something does or doesn't fit within its guardrails plus social consensus. (And as you know, consensus is easy to hack. E.g. writing a paper on global cooling gets you fired, so only global warming papers are left. Therefore AI will tell you global warming has no debate.) It also phrases things how it thinks you'll approve of rather than adhering to what it knows to be true. In other words, it knowingly "white lies."A human with a whole brain can look at methodology and see that something might truthfully contradict doctrine.>>539249312Search engines are useful tools to gather studies to read. You just have to read them. All of them. On your own. Otherwise you're just looking at a map to nowhere and fantasizing about the destination. That's what AI helps you to do - to fantasize.
>>539244735>research is wasting 6 hours of your time to look for sources you can cite on a result from your experiment, but which no journal will publish if you don't have a citation for itunfortunately, truewhat's a meme is reading 35 papers in 6 hours though, unless you're in a field like medicine where the papers are baby-tier and even an amateur can easily breeze through them.in 6 hours, you actually read maybe 5-6 at most.if you "read" 35, then you didn't read shit, you just skimmed them to see if you can use them as citation
>>539249312So you can...do exactly what OP's image says?
>>539249396do you know like a scientist or are you guessing like an llm
>>539249396You HOPE. You don't believe, you wish. See the difference?
>>539244735>NEVER have an original thought. EVER!
>>539244735yet they seethe when i did my own research and chose not to take the clot shot.
>>539249320The reason llms are just guessing is because they are only trained on language. If they had a more holistic training database, like humans do when they are allowed to play as kids, then this would change.
>>539244735>research is spending 6 hours misreading 35 papers that can't be reproduced, so you can write one sentence containing 2 references and be wrongDeath to popscience.
>>539244735>research is reading 33 papers that disagree with you to find and reference the two that doPOWERFUL
>>53924473535 papers in 6 hours sounds like social science or psychology (also social science)
>>539249694>>539249731I know like a scientist because I performed the experiment myself and it has been fixed.
I clicked on this thread thinking it might be interesting but there's absolutely nothing here worth reading.
>>539249396Fixed how? By hard-coding an exception?>>539249853A machine capable of that would be an actual AI. If we ever reach that point then most of humanity's current problems will disappear.
I'm smart and knowledgeable enough that I can write a paper and easily find the references I need afterward.
>>539250323I don't know how they fixed it but it's fixed and the idea that AI can code properly and do math but can't calculate analytically is just ignorant.
>>539250442lmao
>>539250507>it's fixed What is fixed? The cause or the symptom?
>>539250323The only difference between "a machine like that" and an llm is the training data.
>>539250538It guesses linguistic answers now but also is able to translate lignuistic questions into calculable algorithms to come up with actual answers instead of guesses.
people that really love science do it because they love it, but you wouldn't know that, OPyou shouldn't enter science for money or prestige
>>539244735You have any idea how many times you have to sit on a toilet to read 35 papers?
>>539250809As far as we know, humanity has never taught an animal to think critically. Which suggests that a device as powerful as the human brain is required. Current tech is incapable of matching the human brain.
This should serve as a lesson to Pollock who expect rapid and lasting solutions for big problems. Everything good takes tume. And that presumes an attention span. Most of you have an attention span such that you could let read a single book.
