I find the comparison with CERN (made by the Guardian article; CERN is not mentioned in the open letter) misleading and somewhat troubling.
As of 2016 (last year for which there are figures; new ones should be out in a couple of weeks) CERN had 82 research physicists on staff, out of a total headcount of 2531 [1].
The vast majority of researchers working at CERN are associates (982), users and visitors from other institutions, mainly universities in member countries. CERN's primary role is to provide shared facilities (accelerators and detectors) which are too expensive for individual members, not to employ researchers.
The equivalent in machine learning would be shared computing, data collection & labeling resources capable of competing with what Google, Facebook etc have to offer. A state-led initiative on shared computing resources might be more cost-effective than buying time on Google Cloud, but the main thing could well turn out to be pooling data from national databases. Yay?
As of 2016 (last year for which there are figures; new ones should be out in a couple of weeks) CERN had 82 research physicists on staff, out of a total headcount of 2531 [1].
The vast majority of researchers working at CERN are associates (982), users and visitors from other institutions, mainly universities in member countries. CERN's primary role is to provide shared facilities (accelerators and detectors) which are too expensive for individual members, not to employ researchers.
The equivalent in machine learning would be shared computing, data collection & labeling resources capable of competing with what Google, Facebook etc have to offer. A state-led initiative on shared computing resources might be more cost-effective than buying time on Google Cloud, but the main thing could well turn out to be pooling data from national databases. Yay?
[1] https://cds.cern.ch/collection/CERN%20Annual%20Personnel%20S...