4 ms·

> 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

> 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.

danny_codes 2y ago
I wouldn’t trust an LLM to do anything compliance related. Sounds like a recipe for a lawsuit
- teleforce 2y ago
  > cheaper cost using automation, reasoning and decision making with minimum number of expert staff in the loop for verification and validation.
  - javagram 2y ago
    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 2y ago
      > 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.
ModernMech 2y ago
> 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 2y ago
  > 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 2y ago
    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 2y ago
      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.
      - teleforce 2y ago

danny_codes 2y ago

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

teleforce 2y ago
> cheaper cost using automation, reasoning and decision making with minimum number of expert staff in the loop for verification and validation.
- javagram 2y ago
  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 2y ago
    > 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.

ModernMech 2y ago

> 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 2y ago
> 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 2y ago
  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 2y ago
    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.
    - teleforce 2y ago