3 ms·
If it's a science "experiment", his strategy would have to be revealed so you can reproduce it. Names of people participating in drug trials is not required to
by altcognito 15y ago
If it's a science "experiment", his strategy would have to be revealed so you can reproduce it. Names of people participating in drug trials is not required to reproduce an experiment. In principal this makes it different from not publishing the names of people who participated in drug trials.
All he has "proven" is that a certain subset of people can be conned into typing something into at terminal. I don't get the significance. For all I know, he's choosing his target.
- Eliezer 15y agoThe targets are, indeed, selected, by the criterion "You believe not even a transhuman AI could get you to let it out of the box."
- dsr_ 15y agoI might believe a transhuman AI could convince me; I am not convinced that any human can emulate a transhuman AI well enough to do so. Would you say that your winning strategies involved thinking transhumanly (perhaps in non-realtime, a la Vinge's Mailman)?
- Eliezer 15y agoObviously no, essentially by definition.