On secrecy : while I think secrecy on capability research is probably bad because it creates an arm race, winner take all mindset that will not let anyone pause to think when the moment to plug in the actual AGI comes, I think that secrecy for alignement research is just crazy. Open publication = easier progress, easier for outsiders to contribute, easier for orgs that do capability to use the results from alignement.
On recruiting good people : maybe try to recruit bright young people on scholarships/PhD grants and teach them rather than trying to only hire people who are ready to work for free for years before getting hired ? (epistemic status : obviously this is only hearsay from the Internet, may likely not represent the real recruitment process.
On secrecy : while I think secrecy on capability research is probably bad because it creates an arm race, winner take all mindset that will not let anyone pause to think when the moment to plug in the actual AGI comes, I think that secrecy for alignement research is just crazy. Open publication = easier progress, easier for outsiders to contribute, easier for orgs that do capability to use the results from alignement.
On recruiting good people : maybe try to recruit bright young people on scholarships/PhD grants and teach them rather than trying to only hire people who are ready to work for free for years before getting hired ?
(epistemic status : obviously this is only hearsay from the Internet, may likely not represent the real recruitment process.