Given that AI is able to reduce the intellectual labor needed to produce research, and assuming it will get better, does it mean that science research (or R&D and adjacent) could become a low-skill job? Like how weaving also became a low-skill job?
I mean a lot of academia is underfunded, overworks most researchers, often even the top people in the field, and is for the most part pretty low wage, and it can often be very difficult to build a career in many fields. I’d say they are pretty proletarianized in many ways already.
quite possibly but it’s too early to tell rn imo
Science is a high reliability field. If you fuck up any part of a study, ideally it’s torn apart in peer review and your decade of university is for nothing. I think that limits the reach of AI. If you hallucinate any part of your study it isn’t replicable, and you can’t tell if something is hallucinated unless someone with a decade of university is obsessively looking over every scrap of data.
There were articles not that long ago saying a sizeable portion of reviewers were overworked, underpaid, and lied about actually reviewing submissions.
Every element of science is underfunded and overworked. It just isn’t a field where you can bullshit without consequences because nature will test your work even if the institution doesn’t check it first. You invest so much time and money to get established in the field and your professional reputation is scrutinised by a whole media ecosystem outside of work. Plagiarists and charlatans are already villains youtubers have made careers out of attacking. Other teams are citing your research when conducting their own primary research in entirely different fields, and if their work fails they have to explain why. The public gets daily emails from journal aggregators listing new studies, and they’re trained to spot AI slop in all of its other contexts.
That’s too much risk for me. Every word and picture attached to my legal name, Dr. Doctor Richard Richard Richard “Dick” Richard “Rich” Richard “Moneyman” Richard Pennis, is very intentional and backed up by everything else in that work.
The labor aristocracy is a funny thing. They (me included) are untouchable, until they aren’t.
It might just become the same as anything: they won’t be the bench scientist, that do experiments they came up with, have that eureka moment and all the romanticism. Some (from PhDs in Physics and Math that I personally know) have romanticized so much the things that are in that “I just need time, pen and paper to do research”, instead of trying to learn a little bit of programming, try some scenarios, or even some formal proof stuff. They irk themselves of computers, as it feels below them. “That’s for the hoi polloi”, they might think.
But, alas, such awful future came. Not one of automation so we can do more, better, on our own terms, and guide the tech, exploit it to our advantage and betterment of the species. No, you become an “agent swarm” wrangler, needing to work on the machine’s pace, lest you are out of your grant. They are frozen, after all.
To summarize: they got the textile industry treatment, a century and some later, but instead of doing something material, are romanticizing and becoming luddites. They do not see themselves as proles, but they ought to.
(I do know, too, that in STEM, there’s a huge amount of reactionaries, thinking that they are exclusively technical, science is newtral, as well as tech, and humanities are, too, below them).
I highly doubt it. “AI” isn’t a magic bullet. If you don’t understand the underlying science, you will give it bad prompts and be unable to recognize when it produces faulty outputs.
IMO, every scientific institution will change the rules. Science can’t be a low-skill job because it’s applied to the population, which is what R&D is all about, and that can’t be taken lightly. However, the science has learned to work for purely capitalist purposes, and that has led to a huge amount of bogus research, without any AI involvement. Now, can you imagine how much fake research will come out with AI? Scientific journals, along with the professional ethics of each of the scientific researchers, will be largely responsible for these events and will therefore also play an important role in the path towards junk research.
AI is able to write computer programs and solve math problems well but that doesn’t mean that AI is good at science generally. There is something about the nature of the language of math and programming that lends them to be amenable to LLMs. But I doubt you will see the same level of efficiency in something things like real world engineering, sociology, maybe even theoretical physics etc.
Depends on what you mean by proletarianized.
does it mean that science research (or R&D and adjacent) could become a low-skill job?
This reads a little like science is something that can be solved, which it of course isn’t. Reality is extremely complex, our goal is to understand it. Today, science performed by humans is highly specialized and complex, thus even if the barrier to “producing” new science would be lower, the actual models, theories or explanations wouldn’t be less complex – on the contrary they would probably be much more complex, requiring even more education, specialization and skill of the human operator or reviewer.
So framing it as a future “low-skill” endeavor makes little sense. The question is rather how the neoliberal system of creating science will be severely limited in how it utilizes these tools and their acceleration of the existing publish-or-perish regime that currently rules academia.
I don’t think that the neoliberal system will cope with AI tbh. Like other comrade said here, Western academia is probably gonna get flooded by fake publications given that it already is flooded by non-reproducible results.
I don’t believe that AI will turn science into a low-skill job, but I was curious about the perspectives of comrades here.
