>Scientists who engage in AI-augmented research publish 3.02 times more papers, receive 4.84 times more citations and become research project leaders 1.37 years earlier than those who do not. By contrast, AI adoption shrinks the collective volume of scientific topics studied by 4.63% and decreases scientists’ engagement with one another by 22%.https://www.nature.com/articles/s41586-025-09922-yWhat do you make of this? I've been watching my mom go from a mediocre researcher (by her own admission) to a productive one with the help of AI, and I'm curious if you've seen similar situations in your neck of the woods.
>>16893355>The value of a researcher comes from his publication ration and citation ratioCan someone tell me what science was supposed to be about again?
>>16893359This was a theoretical physics paper the core idea of which was AI-generated, to the extent that Steve Hsu regards GPT as the main contributor. You should read about his process for getting results. AI is a useful tool even at the frontier of physics.https://arxiv.org/abs/2511.15935
>>16893355This is bullshit, AI is useless.
>>16893363To what extent is this AI-enabled physics paper useless in your view?https://arxiv.org/abs/2511.15935
>>16893355But nobody likes it when you use AI.All the journals and online publishers say if you use AI you are not welcome.
>>16893366Arxiv is a bunch of rats. They rejected my paper yesterday, calling it not original enough.
>>16893359They/them have strayed away from the light of Christ and become nothing but beggars looking for crumbs falling from the table of the oligarchs.
ai slopience feedback cycles will be a horror show.if you thought fraud and non replicability were bad already it will only get worse.
>>16893366>Hey chatgpt rewrite these disjointed notes as a more coherent paragraph >Hey chatgpt find the research papaer on X>"AI augmented research"
AI is pretty awesome at literature search and doing paper digests.
>>16893355>3 times more fake papers>5 times more citations by chatbots citing their own slop>at least you get to lead "projects" with worthless members who also use chatbots to shit out results>everyone knows the game, no one bothers to engage with slop from unaffiliated "researchers"The only possible outcome for the publishing mills. Why use people to shit out junk science when you can use bots?
>ai written papers>ai peer review>ai editor>ai journaland all to support the implementation of some political policy
>>16893609>doing paper digests.If you ever manage to form your own original thoughts (unlikely) try making a chatbot doing a "digest" on them and see how much you like the result. Retard.
>>16893355AI has its uses, especially if it's being used by people who know what they're doing, but it also produces a helluva lot of slop and the uninitiated often don't recognize the difference.It's also useless for the most important part of the scientific process: Experimentation. Automation can certainly streamline parts of data collection and analysis, but at the end of the day you still need people to design experiments, fabricate parts, build apparatuses, collect data, and interpret results, and AI is still a long, LONG ways off from being able to be trusted with automating chunks of that process.
I think that using AI for informations/research is not reliable at all.AI isn't competent enough to filter informations and understand some nuances only a proper competent human could.But I do think that AI is usefull when it comes to save a lot of time.With the proper prompt and using your own filtered informations/research and giving it to AI , it can do wonder.
>>16893361>You should read about his process for getting results. Honestly, these attempts to normalize AI are pathetic. I don’t know how Steve Hsu did it, but the stuff that Tao did was literally consulting a magic 8 ball for random ideas. Ideas that he would come up himself naturally. I don’t like the word. But it’s a fucking psyop. Either they did it as a favor or they were bribed/blackmailed to praise AI in public. In Tao’s case, it’s probably both. Old colleagues in Microsoft Research and microsoft money. The whole thing is disgusting.
>>16893361>AI is a useful tool even at the frontier of physics.By which you mean it's a useful tool for generating more of the same kind of meaningless pseudoscience "the frontier of physics" has become infamous for.
>>16893645Everything related to AI use comes down to a simple question: Does the person using it understand what they're researching enough both verify and effectively apply generated content?There's a very, very broad line between the people who use AI to do literature searches and then take the time to verify all the papers it finds and the summaries it produces, or the people who use it to generate code and then actually parse through what it generates and confirms what each line does and how it fits together, and so on... and the schizos who just prompt AI to shit out some garbage numerological music of the spheres spiritualist mumbo jumbo quackery and then dump it on /sci/ or reddit or wherever and declare that they've "solved science".
>>16893701No. It comes down to the question of whether or not they understand how this technology works. And the answer is almost always 'no'. Because they did, they wouldn't be using it for "science".
I think it has value as an extended search function to pull out data that might be relevant but I don't think it's capable of critical insight
>>16893742This. It's moderately useful for searching for more specific things than most search engines (even those specifically made for scientific literature), but you've nearly always got to fact-check the results.
