[teleforce]

> one clear driver is continually increasing rules, regulations, and compliance, along with fears of audits and lawsuits

I think this a gap that can be easily and fittingly addressed by explainable AI (XAI) hopefully with much cheaper cost using automation, reasoning and decision making with minimum number of expert staff in the loop for verification and validation.

I've got the feeling that Elon proposed DOGE as a trojan horse for doing this sneakily:

1) Reduced the budget to make govt more efficient so staff number reduction is inevitable

2) Sell and provide XAI based solutions for regulatory compliance, etc (accidentally his AI company name is xAI)

3) Repeat these with many govt's organization, research, academic institutions

4) Profit!

But apparently the US research universities like UPenn did not get the memo and cut the number of graduate research students instead of the admin staff.

[ModernMech]

> But apparently the US research universities like UPenn did not get the memo and cut the number of graduate research students instead of the admin staff.

If you reduce the number of staff, the people who are going to hurt first are the graduate researchers. I run my lab with a whole host of college and department staff who make all of our jobs easier. If you cut them, their jobs are going to fall to professors, and if they have to do more admin work, graduate teaching and research assistants are going to get more shit work, and also there's going to be fewer than them.

For instance we have a whole office that help us get our research funded. These people are "bureaucratic administrative overhead", but they make everyone's job easier by providing a centralized resource for this particular problem. Get rid of them an you can save millions of dollars in salaries, but you're going to lose more than that in lost contracts and professor/student productivity. This would mean students probably would get cut anyway, so they're making the smart move of supporting only the students they can, and not leaving anyone out to dry.

[coliveira]

> If you cut them, their jobs are going to fall to professors

yes, then you hire more professors, instead of hiring more staff! Funny how people don't seem to realize the obvious.

[ModernMech]

I thought the purpose of this was to reduce waste. Firing a low cost administrator and replacing them with N highly-trained (and higher cost) Ph.D.s is not efficient use of taxpayer dollars.

[coliveira]

The goal is to have more researchers that will do high impact research. You cannot do that if you take all the resources and spend on managers and administrative staff.

[ModernMech]

If that's the goal, firing the administrative staff will have exactly the opposite effect. Administrative staff and managers free researchers to do research. If administrative staff and managers are fired, researchers will be administrating and managing instead of researching.

Getting rid of administrators doesn't obviate the need to administrate. It has to be done, so we do it efficiently using shared resources, which brings economies of scale -- that efficiency Musk keeps talking about. What you're arguing for is increasing waste so everyone has less time to do critical work.

Here's an analogy:

To support the roof of a house, you need a few support beams. To support the roof of a skyscraper, you need many more support beams. You can't support the roof of a skyscraper with the number of support beams that support the roof of a house.

University research started as a house, but now it's a sky scraper. You're coming into the skyscraper saying there's too many beams, but you're judging by house standards. Maybe there are, but most of them were put there for good, well-considered reasons; as a layman you have no idea which are load bearing, so if you come knocking them down you endanger the whole tower. Which is a shame because it's gotten really really tall - taller than any other tower in history - so toppling it because you don't understand it would be a huge loss for everyone.

[teleforce]

> The goal is to have more researchers that will do high impact research

Given that my comments are downvoted like crazy, I've got the feeling that the US university including the Professors (tenured) are missing the forest from the trees regarding this issue.

I once asked a senior and prominent US Professor regarding their multi-million dollars grant for single project that can be easily spent on multi-project with similar or higher impact in other countries. His answer was they have to spent a lot on students, and now I know the truth that most of the money are going to the research managers and admin staff, what as waste.

[ModernMech]

> they have to spent a lot on students

This is a good thing. It's expensive to support a Ph.D. student in America; it's a lot cheaper if you're in a country with lower cost of living. But as a researcher, you want to do research in an expensive area because it means you'll be around other smart people and lots of resources.

At the end of the day tho, despite all its flaws, this system is a winner; US produces the most research, is home to the best universities, and students from around the world dream of studying in America. We can make improvements, but the need for a rewrite of the system is greatly overstated. Other countries wish they had our problems.

[danny_codes]

I wouldn’t trust an LLM to do anything compliance related. Sounds like a recipe for a lawsuit

[teleforce]

> cheaper cost using automation, reasoning and decision making with minimum number of expert staff in the loop for verification and validation.

[javagram]

Verification and validation of LLM output in this context would mean doing all the same research, training etc done today for human staff and then comparing the results line by line. It would actually take more time. How do you know if the LLM failed to apply one of hundreds of rules from a procedure unless you have a human trained on it who has also examined every relevant document and artifact from the process?

[teleforce]

> one clear driver is continually increasing rules, regulations, and compliance, along with fears of audits and lawsuits

As mentioned by the GP posts the main problem is the increasing rules, regulations and compliance need to be processed the admin staff not the research contributions itself (these invention and innovation parts are performed by the graduate students and professors who are getting cuts by the limited budget).

This AI based system will include (not limited to) LLM with RAG (with relevants documents) that can perform the work of the tens if not hundreds jobs of the admin staff. The agent AI can also include rule based expert system for assessment of the procedures. It will be much faster than human can ever be with the on-demand AWS scale scaling (pardon the pun).

Ultimately it will need only a few expert admin staff for the compliance validation and compliance instead tens of hundreds as typical now in research organizations. The AI based system will even get better over time due to this RLHF and expert human-in-the-loop arrangement.