4 ms·

Correlation does not imply causation. How do these AI systems account for causation? It's the fundamental issue I have with the penetration of these sorta "AI

Correlation does not imply causation. How do these AI systems account for causation? It's the fundamental issue I have with the penetration of these sorta "AI" systems in daily life: from what I understand about the basic maths, they're ultimately all seeking correlation and real-world data used to improve models seems to achieve verification, not validation. I'm guessing I'm not the only person who reads this article and experiences a visceral reaction. This is what I would call "ugly" tech.

kmlx 7y ago
this product is about prevention from my understanding. not about calling the thought crime police. i think it’s quite great tech, definitely a huge need in the market for this sort of prevention mechanism.
- kartikkumar 7y ago
  The risk of falsely predicting that someone is going to shoplift, based on models that are driven by correlation, is not a problem? How is it not about policing thought? If you present physical attributes/symptoms of someone who is thinking about shoplifting (nervousness, anxiety, fidgetiness, etc., as mentioned in the article), the models can classify you are "about to shoplift". That's pretty close to thought policing if you ask me.
doublekill 7y ago
These systems are interested in probabilities, not in finding the causes of shoplifting. If not correlation, then what else?
- kartikkumar 7y ago
  How is a validated chain of causation not important for these systems to arrive at ethical predictions? The causation gap for these models presents a really terrible scenario. Without causation, these models are capable of correlating things like race, gender, etc. with outcomes like shoplifting.
  - doublekill 7y ago
    Because it is not a social science paper or a government policy (smoking causes cancer, so we should forbid advertising smoking to teenagers). You do not combat shoplifting by knowing the causes of shoplifting (being unethical causes shoplifting, now what?). Nobody infringes on your rights, if a system pings you as suspicious and so a security guard is alerted and watches you more closely. There are no false positive consequences here: either they find that can of Red Bull on your person without a paid receipt, or they don't. You won't get banned from the store, because the system inadvertently deemed you suspicious. A whitebox statistical model still can provide the factors that contributed to its prediction, without resorting to causal inference. I do not think a certain race, gender, age, income causes shoplifting, but they sure are correlated, and effective at finding shoplifters or tax fraudsters. I deem the act of shoplifting more unethical than the act of watching you on a camera feed. This already happens anyway, these systems just manage attention better.
dxbydt 7y ago
>Correlation does not imply causation. i think we should stop trotting out this tired nugget without understanding how sparse the alternative is. Even within the academic stats community, there’s a fair amount of handwringing that rho is honestly all we have. fpci is one of the most dispiriting theorems in statistics. we get by with rubin models & sem, but these are pretty hokey and outside of social sciences you get laughed at if you say ran an sem. you can regress student score on study time & get a positive coefficient, but then you are setting yourself up - why did you get that positive coefficient ? well it means if he studied longer he would have scored higher. yeah but then why did he not study longer ? the fuck i know! oh wait a minute let’s do sem! ok we can’t observe depression so that’s a latent. and we can’t observe teenage angst so another latent. soon your friendly neighborhood psychologist thinks up half a dozen latent causes. put the whole thing into the sem mixer grinder with score as response. yay sem software gave me significance! ok so depressed teenagers will score poorly on test because they won’t study for it...but wait a minute why are they depressed- oh lets do another sem throw it divorce rate & obesity & lack of religious attendance & regress again....this is when the statistician gets up & says fuck all you assholes you don’t have anything to establish any of these things, you are just grossly misusing linear regression. then the social scientists respond well we don’t need your math kind anyways, we have sem software, its latent turtles all the way down. so what if beta hat is xprime x inverse xprime y? can’t publish a psych paper with just that. gotta have my latent goodies. causal inference ftw! god sometimes this whole field makes me hurl.

kmlx 7y ago

this product is about prevention from my understanding. not about calling the thought crime police. i think it’s quite great tech, definitely a huge need in the market for this sort of prevention mechanism.

kartikkumar 7y ago
The risk of falsely predicting that someone is going to shoplift, based on models that are driven by correlation, is not a problem? How is it not about policing thought? If you present physical attributes/symptoms of someone who is thinking about shoplifting (nervousness, anxiety, fidgetiness, etc., as mentioned in the article), the models can classify you are "about to shoplift". That's pretty close to thought policing if you ask me.

doublekill 7y ago

These systems are interested in probabilities, not in finding the causes of shoplifting. If not correlation, then what else?

kartikkumar 7y ago
How is a validated chain of causation not important for these systems to arrive at ethical predictions? The causation gap for these models presents a really terrible scenario. Without causation, these models are capable of correlating things like race, gender, etc. with outcomes like shoplifting.
- doublekill 7y ago
  Because it is not a social science paper or a government policy (smoking causes cancer, so we should forbid advertising smoking to teenagers). You do not combat shoplifting by knowing the causes of shoplifting (being unethical causes shoplifting, now what?). Nobody infringes on your rights, if a system pings you as suspicious and so a security guard is alerted and watches you more closely. There are no false positive consequences here: either they find that can of Red Bull on your person without a paid receipt, or they don't. You won't get banned from the store, because the system inadvertently deemed you suspicious. A whitebox statistical model still can provide the factors that contributed to its prediction, without resorting to causal inference. I do not think a certain race, gender, age, income causes shoplifting, but they sure are correlated, and effective at finding shoplifters or tax fraudsters. I deem the act of shoplifting more unethical than the act of watching you on a camera feed. This already happens anyway, these systems just manage attention better.

dxbydt 7y ago

>Correlation does not imply causation. i think we should stop trotting out this tired nugget without understanding how sparse the alternative is. Even within the academic stats community, there’s a fair amount of handwringing that rho is honestly all we have. fpci is one of the most dispiriting theorems in statistics. we get by with rubin models & sem, but these are pretty hokey and outside of social sciences you get laughed at if you say ran an sem. you can regress student score on study time & get a positive coefficient, but then you are setting yourself up - why did you get that positive coefficient ? well it means if he studied longer he would have scored higher. yeah but then why did he not study longer ? the fuck i know! oh wait a minute let’s do sem! ok we can’t observe depression so that’s a latent. and we can’t observe teenage angst so another latent. soon your friendly neighborhood psychologist thinks up half a dozen latent causes. put the whole thing into the sem mixer grinder with score as response. yay sem software gave me significance! ok so depressed teenagers will score poorly on test because they won’t study for it...but wait a minute why are they depressed- oh lets do another sem throw it divorce rate & obesity & lack of religious attendance & regress again....this is when the statistician gets up & says fuck all you assholes you don’t have anything to establish any of these things, you are just grossly misusing linear regression. then the social scientists respond well we don’t need your math kind anyways, we have sem software, its latent turtles all the way down. so what if beta hat is xprime x inverse xprime y? can’t publish a psych paper with just that. gotta have my latent goodies. causal inference ftw! god sometimes this whole field makes me hurl.