I will have something to say about the guy who murdered two people in Minnesota, tried to murder two more, and intended to murder dozens, but I'm waiting until I'm absolutely sure what he was all about. The corporate media are being very reticent to discuss his motives, which may mostly be fear of offending the people who share his ideology, but it might be that it isn't exactly as it appears. Note that MAGAts are climbing all over each other to claim that he is actually some sort of disgruntled radical leftist, which is a blatant and insane lie, no doubt about that. But I'm better than them, so I'm not jumping to any conclusions. -- C
Part 4 of Gioia's essay is titled "Funding for science and tech research is disappearing in every sphere and sector." That's a pretty sweeping claim? Let's see if it's true. First I'll just quote the entire section:
The whole technology and science power structure requires research—and somebody must pay for it. But, in very short order, the major sources of funding have dried up.
This is more than just a change in government policy. Even the huge corporations that fund their own research programs are now investing in AI data centers, not scientists. Somebody should measure this, but I’m confident that the shift from human-driven R&D projects to capital equipment investing is enormous. That’s why Meta is preparing for layoffs. That’s why Microsoft is getting rid of software jobs. That’s why Google is elminating people.
I need to emphasize that this is NOT a short term economic trend. Even the most successful tech companies are losing their appetite for human-driven research projects.
Well, in the first place, the major source of funding for science, per se -- basic science, trying to understand how the universe and organisms and human brains work -- is government. Wealthier nations fund most of it, for the obvious reason that they have the money. It is true that the U.S., in the past 5 months, has cut back drastically on its funding for scientific research, but that obviously is indeed, contrary to Gioia, a short term trend, and it's quite unpopular. European countries and Japan, and even China, are now trying to aggressively recruit U.S. scientists.
Microsoft, Meta and Apple have never funded very much scientific research. They have funded technological applications of scientific knowledge, which is not the same thing. You can learn some science along the way, to be sure, and discover some anomalies that suggest research questions to real scientists, but the objective of the research is not to learn fundamental truths about the universe but to make products that can be sold at a profit. Those corporations are still trying to do that, but like every other industry, when they have a chance to profitably substitute capital for labor, they'll do that. They're probably making a mistake thinking that so-called Artificial Intelligence (a grotesque misnomer) is where they'll make the big bucks in the future, but just because they're doing that with fewer computer programmers than they formerly employed doesn't mean they aren't engaged in technological research.
Basic scientific research enterprises used to employ a lot of humans, whose job title was "computer." (They were mostly women.) Now that's no longer a job title because machines do it. And they can do computations that humans could never do. But just because fewer people are employed to do computation doesn't mean that computers meant less scientific research was happening.
Gioia seems to think that employment of computer programmers is the measure of investment in scientific research. That doesn't mean we don't have a problem, it just means that, once again, Gioia has misidentified it.
3 comments:
Something has been bothering me lately. As I understand it ML (Machine Learning = "AI") algorithms using LLM's (Large Language Models - Not the only technique used by ML) depend on collecting as much of the text "out there" as it can (Plus other stuff?) and "trains" on it. If these systems really catch on then a lot of the "stuff out there" will be generated
by themselves. As they update their "knowledge base" (Sorry about all the "'s and ()'s.) won't they then be training on their own output? I'm told that echo chambers aren't a good source to learn from.
In addition to attacking science Trump seems to be targeting a number of our nation's strengths - Universities, our well-functioning Civil Service supporting essential needs, our independent press, and keeping ourselves nimble by welcoming a steady stream of people with the courage, self-confidence, and optimism required to emigrate from the world they know.
It seems like a pretty effective blueprint for an enemy to surreptitiously destroy our nation from within. How will this MAGA?
Hey CP, regarding point one, as I understand it this is a concern. If they start digesting their own output LLMs will deteriorate. I assume the companies are aware of this and have some plan to address it, but if the concept is to remain viable humans will have to continue to produce most of the available text. That does strike me as a substantial limitation. In other words, they can't actually think or create, they can only plagiarize.
As for your second point, yes indeed, and that is the classic fascist playbook. The first thing the Nazis did when they invaded Poland was not to go after the Jews, but rather the university professors, journalists, and civil servants. (Of course there was overlap.) And obviously they did the same in Germany before they started on their campaign of conquest. Dictators don't want any independent sources of truth.
Post a Comment