Yesterday, National Academy of Sciences President Marcia McNutt delivered her last annual State of the Sciences Address. Overall the talk basically calls us to adapt to the new reality that industrial and foundation support for research has taken a far larger role in academic research. I fear we may be losing the broad academic independence and exploration that has made our universities the envy of the world.
Government funding for research has become more challenging to receive, more bureaucratic and more political. The US Office of Management and Budget has proposed new regulations that would require all grant funding to go through political review.
McNutt presented this graph showing the seemingly exponential growth in research requirements from zero in 1990 (when I got my first grant) to well over three hundred today. Researchers now spend close to half their time on regulatory compliance.
 |
| Increase in regulatory requirements |
McNutt mentioned six specific issues.
- Reimagine connections between universities and industry
- Realign the academic reward system
- Meet the needs of the STEM workforce
- Reduce research regulations (as in the graph above)
- Increase the rate of innovation with the help of automation in shared facilities
- Tackle Big Challenges
For the first, she highlighted U Washington CS where a large percentage of their faculty had half-time appointments in industry. I understand why this is necessary from both sides, but that doesn't mean it's a good thing. Academics should not have two masters. We become academics to choose our own research directions and to prepare the next generation, both more challenging when you spend half your time connected with a company.
Even for the "big challenges" she specifically mentioned two foundation-driven projects.
McNutt has hopes that government research can be less bureaucratic and willing to take bigger risks. Outside of theoretical areas, CS conferences now have heavy industrial representation. I worry that we bend our research to fit the needs of corporations and foundations funded by wealthy donors. Welcome to the new world.
In current era of AI, no one is talking about time complexity. Everyone just want more data and more compute.
ReplyDeleteIt literally says CLOGR on the chart
ReplyDeleteLook up BitNet1.58
ReplyDeleteVery valid concerns. I see it as a part of the general pattern of our sociopolitical institutional infrastructure in the US being under immense pressure over the recent years. It needs a reflection on the roots of public discontent with our institutions.
ReplyDeleteAdditionally, it might be time for universities to try to get a share of any value created as a result of their research when it is commercialized. We can keep the non-commercial use open source and free and even small commercial use free but charge for usage of our research by big corporations through licensing agreements and reversing the direction of the relationship. That should give university enough money to then redistribute across other fields.
Industry funded research, it can similarly evolve to build new non-governmental institutions which gather donations from industry and carry out NSF and NIH like granting process independent of the government politics.
What I worry also about is that many frontier labs in AI have limited what they publish to keep an edge in front of other players. It is a very bad idea, cause you can easily hire researchers from other labs and they would know it, but the general public would not, slowing the research. I also worry that Chinese institutions with their more open approach to AI might end up becoming the dominant force and leave the US institutions behind in research.