One man’s ‘warning shot’ is just another man’s “easily patched minor bug of no importance if you aren’t anthropomorphizing irrationally”, because by definition, in a warning shot, nothing bad happened that time. (If something had, it wouldn’t be a ‘warning shot’, it’d just be a ‘shot’ or ‘disaster’.
I agree that “warning shot” isn’t a good term for this, but then why not just talk about “non-catastrophic, recoverable accident” or something? Clearly those things do sometimes happen, and there is sometimes a significant response going beyond “we can just patch that quickly”. For example:
The Three Mile Island accident led to major changes in US nuclear safety regulations and public perceptions of nuclear energy
9/11 led to the creation of the DHS, the Patriot Act, and 1-2 overseas invasions
The Chicago Tylenol murders killed only 7 but led to the development and use of tamper-resistant/evident packaging for pharmaceuticals
The Italian COVID outbreak of Feb/Mar 2020 arguably triggered widespread lockdowns and other (admittedly mostly incompetent) major efforts across the public and private sectors in NA/Europe
I think one point you’re making is that some incidents that arguably should cause people to take action (e.g., Sydney), don’t, because they don’t look serious or don’t cause serious damage. I think that’s true, but I also thought that’s not the type of thing most people have in mind when talking about “warning shots”. (I guess that’s one reason why it’s a bad term.)
I guess a crux here is whether we will get incidents involving AI that (1) cause major damage (hundreds of lives or billions of dollars), (2) are known to the general public or key decision makers, (3) can be clearly causally traced to an AI, and (4) happen early enough that there is space to respond appropriately. I think it’s pretty plausible that there’ll be such incidents, but maybe you disagree. I also think that if such incidents happen it’s highly likely that there’ll be a forceful response (though it could still be an incompetent forceful response).
I think all of your examples are excellent demonstrations of why there is no natural kind there, and they are defined solely in retrospect, because in each case there are many other incidents, often much more serious, which however triggered no response or are now so obscure you might not even know of them.
Three Mile Island: no one died, unlike at least 8 other more serious nuclear accidents like (not to mention, Chernobyl or Fukushima). Why did that trigger such a hysterical backlash?
(The fact that people are reacting with shock and bafflement that “Amazon reopened Three Mile Island just to power a new AI datacenter” gives you an idea of how deeply wrong & misinformed the popular reaction to Three Mile was.)
9/11: had Gore been elected, most of that would not have happened, in part because it was completely insane to invade Iraq. (And this was a position I held at the time while the debate was going on, so this was something that was 100% knowable at the time, despite all of the post hoc excuses about how ‘oh we didn’t know Chalabi was unreliable’, and was the single biggest blackpill about politics in my life.) The reaction was wildly disproportionate and irrelevant, particularly given how little response other major incidents received—we didn’t invade anywhere or even do that much in response to, say, the first World Trade attack by Al Qaeda, or in other terrorist incidents with enormous body counts like the previous record holder, Oklahoma City.
Chicago Tylenol: provoked a reaction but what about, say, the anthrax attacks which killed almost as many, injured many more, and were far more tractable given that the anthrax turned out to be ‘coming from inside the house’? (Meanwhile, the British terrorist group Dark Harvest Commando achieved its policy goals through anthrax without ever actually using weaponized anthrax, and only using some contaminated dirt.) How about the 2018 strawberry poisonings? That lead to any major changes? Or you remember that time an American cult poisoned 700+ people in a series of terrorist attacks, what happened after that? No, of course not, why would you—because nothing much happened afterwards.
the Italian COVID outbreak helped convince the West to do some stuff… in 2020. Of course, we then spent the next 3 years after that not doing lockdowns or any of those other ‘major efforts’ even as vastly more people died. (It couldn’t even convince the CCP to vaccinate China adequately before its Zero COVID inevitably collapsed, killing millions over the following months—sums so vast that you could lose the Italian outbreak in a rounding error on it.)
Thus, your cases demonstrate my claim: warning shots are not ‘non-catastrophic recoverable accidents’ - because those happen without anyone much caring. The examples you give of successful warning shots are merely those accidents which happened, for generally unpredictable, idiosyncratic, unreplicable reasons, to escape their usual reference class and become a big deal, The Current Thing, now treated as ‘warning shots’. Thus, a warning shot is simply anything which convinces people it was a warning shot, and defined post hoc. Any relationship to objective factors like morbidity or fatality count, rather than idiosyncratic details of, say, how some old people in Florida voted, is tenuous at best.
Yeah that makes sense. I think I underestimated the extent to which “warning shots” are largely defined post-hoc, and events in my category (“non-catastrophic, recoverable accident”) don’t really have shared features (or at least features in common that aren’t also there in many events that don’t lead to change).
I agree that “warning shot” isn’t a good term for this, but then why not just talk about “non-catastrophic, recoverable accident” or something? Clearly those things do sometimes happen, and there is sometimes a significant response going beyond “we can just patch that quickly”. For example:
The Three Mile Island accident led to major changes in US nuclear safety regulations and public perceptions of nuclear energy
9/11 led to the creation of the DHS, the Patriot Act, and 1-2 overseas invasions
The Chicago Tylenol murders killed only 7 but led to the development and use of tamper-resistant/evident packaging for pharmaceuticals
The Italian COVID outbreak of Feb/Mar 2020 arguably triggered widespread lockdowns and other (admittedly mostly incompetent) major efforts across the public and private sectors in NA/Europe
I think one point you’re making is that some incidents that arguably should cause people to take action (e.g., Sydney), don’t, because they don’t look serious or don’t cause serious damage. I think that’s true, but I also thought that’s not the type of thing most people have in mind when talking about “warning shots”. (I guess that’s one reason why it’s a bad term.)
I guess a crux here is whether we will get incidents involving AI that (1) cause major damage (hundreds of lives or billions of dollars), (2) are known to the general public or key decision makers, (3) can be clearly causally traced to an AI, and (4) happen early enough that there is space to respond appropriately. I think it’s pretty plausible that there’ll be such incidents, but maybe you disagree. I also think that if such incidents happen it’s highly likely that there’ll be a forceful response (though it could still be an incompetent forceful response).
I think all of your examples are excellent demonstrations of why there is no natural kind there, and they are defined solely in retrospect, because in each case there are many other incidents, often much more serious, which however triggered no response or are now so obscure you might not even know of them.
Three Mile Island: no one died, unlike at least 8 other more serious nuclear accidents like (not to mention, Chernobyl or Fukushima). Why did that trigger such a hysterical backlash?
(The fact that people are reacting with shock and bafflement that “Amazon reopened Three Mile Island just to power a new AI datacenter” gives you an idea of how deeply wrong & misinformed the popular reaction to Three Mile was.)
9/11: had Gore been elected, most of that would not have happened, in part because it was completely insane to invade Iraq. (And this was a position I held at the time while the debate was going on, so this was something that was 100% knowable at the time, despite all of the post hoc excuses about how ‘oh we didn’t know Chalabi was unreliable’, and was the single biggest blackpill about politics in my life.) The reaction was wildly disproportionate and irrelevant, particularly given how little response other major incidents received—we didn’t invade anywhere or even do that much in response to, say, the first World Trade attack by Al Qaeda, or in other terrorist incidents with enormous body counts like the previous record holder, Oklahoma City.
Chicago Tylenol: provoked a reaction but what about, say, the anthrax attacks which killed almost as many, injured many more, and were far more tractable given that the anthrax turned out to be ‘coming from inside the house’? (Meanwhile, the British terrorist group Dark Harvest Commando achieved its policy goals through anthrax without ever actually using weaponized anthrax, and only using some contaminated dirt.) How about the 2018 strawberry poisonings? That lead to any major changes? Or you remember that time an American cult poisoned 700+ people in a series of terrorist attacks, what happened after that? No, of course not, why would you—because nothing much happened afterwards.
the Italian COVID outbreak helped convince the West to do some stuff… in 2020. Of course, we then spent the next 3 years after that not doing lockdowns or any of those other ‘major efforts’ even as vastly more people died. (It couldn’t even convince the CCP to vaccinate China adequately before its Zero COVID inevitably collapsed, killing millions over the following months—sums so vast that you could lose the Italian outbreak in a rounding error on it.)
Thus, your cases demonstrate my claim: warning shots are not ‘non-catastrophic recoverable accidents’ - because those happen without anyone much caring. The examples you give of successful warning shots are merely those accidents which happened, for generally unpredictable, idiosyncratic, unreplicable reasons, to escape their usual reference class and become a big deal, The Current Thing, now treated as ‘warning shots’. Thus, a warning shot is simply anything which convinces people it was a warning shot, and defined post hoc. Any relationship to objective factors like morbidity or fatality count, rather than idiosyncratic details of, say, how some old people in Florida voted, is tenuous at best.
Yeah that makes sense. I think I underestimated the extent to which “warning shots” are largely defined post-hoc, and events in my category (“non-catastrophic, recoverable accident”) don’t really have shared features (or at least features in common that aren’t also there in many events that don’t lead to change).