Sure, but as soon as someone says "what are the odds someone with X features kills himself" - well I didn't invoke the statistical argument did I?
The answer is: it's right within the profile. You don't get to say "what are the odds!?" and then complain about the actual statistics - as noted elsewhere in this thread, the Birthday Paradox[1] is also at play.
What are the odds of any individual whistleblower dying? Who knows. What are the odds of someone, somewhere, describable as a whisteblower dying? Fairly high if there's even a modest number of whistleblowers relative to the method of death (i.e. Boeing has dozens of whistleblower cases going, and OpenAI sheds an employee every other week who writes a critique on their blog about the company).
This same problem turns up with any discussion of vaccines and VAERS. If I simply go and wave my hand over the arm of 1,000 random people then within a year it's virtually guaranteed at least 1 of them will be dead, probably a lot more[2]. Hell, at a standard death rate of 8.1/1000, OpenAI's standing number of employees of 1,700[3] means in any given year it's pretty likely someone will die - and since "worked for OpenAI" is a one-way membership, year over year "former OpenAI employee dies" gets more and more likely.
The answer is: it's right within the profile. You don't get to say "what are the odds!?" and then complain about the actual statistics - as noted elsewhere in this thread, the Birthday Paradox[1] is also at play.
What are the odds of any individual whistleblower dying? Who knows. What are the odds of someone, somewhere, describable as a whisteblower dying? Fairly high if there's even a modest number of whistleblowers relative to the method of death (i.e. Boeing has dozens of whistleblower cases going, and OpenAI sheds an employee every other week who writes a critique on their blog about the company).
This same problem turns up with any discussion of vaccines and VAERS. If I simply go and wave my hand over the arm of 1,000 random people then within a year it's virtually guaranteed at least 1 of them will be dead, probably a lot more[2]. Hell, at a standard death rate of 8.1/1000, OpenAI's standing number of employees of 1,700[3] means in any given year it's pretty likely someone will die - and since "worked for OpenAI" is a one-way membership, year over year "former OpenAI employee dies" gets more and more likely.
[1] https://en.wikipedia.org/wiki/Birthday_problem
[2] https://www.cia.gov/the-world-factbook/field/death-rate/
[3] https://en.wikipedia.org/wiki/OpenAI