Live data from Hacker News

Clearview AI being used to identify Capitol rioters

nytimes.com

71–80 of 179 posts

Re: Clearview AI being used to identify Capitol rioters

#71
post #54
post #37

Earlier quoted context omitted.

Out of curiosity, why do you feel like you're supposed to be supportive of the use of facial recognition in this case?

Identifying terrorists is a positive outcome.

Identifying terrorists and violent criminals is always the argument in favor of using facial recognition for law enforcement.

Re: Clearview AI being used to identify Capitol rioters

#72

Facial Recognition Tech is obviously good if used by competent people: Narrows the number of haystacks to search for needle => Reduced resources required for successful search => More crimes prosecuted. Problem occurs when: * Users of FRT assume all in haystack are needles * Crimes on book must not be universally prosecuted The first part can't be helped. US Police, like most US government jobs, is a rest-and-take-it…

> I want to preserve inequitable outcomes in policing because society is stronger with inequitable outcomes - permits good life for high percentile individuals and constrains operations for low percentile individuals. Thanks for being clear about your perspective. Do you think there's potential for abuse with different rulesets for different people? > Demarcating crime from uncrime is Sorites paradox. I disagree. The…

> Thanks for being clear about your perspective. Do you think there's potential for abuse with different rulesets for different people?

Oh, most certainly. The same structure allows for racial discrimination, which I do not believe is a sensible angle of discrimination: i.e. I think Ben Carson should not be discriminated against for being black. Too high value as a top surgeon.

On the whole I accept it, though, because I don't want Elon Musk prosecuted by the SEC and the instrument that permits both is blunt.

> The measure of a crime is subjective and objective. Subjectively, the victim notices they have been wronged. Objectively, there is a claim by a plaintiff against a defendant. A claim either exists or it does not, there is no sorites.

Indeed. When there is a threshold. However, the costs imposed on society by drug users are dispersed. You can't Categorical Imperative them because some people are not capable enough to handle the responsibility.

Other times the crime is exposure to increased risk: no actual harm may occur. For instance, if you do burnouts on city roads there is little concrete harm, only increased exposure to risk.

It's the same with many things: public drunkenness, drink driving, jaywalking. And society reacts to these by permitting these activities in practice for high-value individuals while proselytizing against them at the same time.

I don't drink-drive but I happily do the other two.

Re: Clearview AI being used to identify Capitol rioters

#73

Facial Recognition Tech is obviously good if used by competent people: Narrows the number of haystacks to search for needle => Reduced resources required for successful search => More crimes prosecuted. Problem occurs when: * Users of FRT assume all in haystack are needles * Crimes on book must not be universally prosecuted The first part can't be helped. US Police, like most US government jobs, is a rest-and-take-it…

That's succinctly put.

I'd like to rob a bank.

I want enough bank robberies to succeed that the fictional movies and books which are an import part of my life to still be plausible.

In general I don't want just anyone to be able to rob a bank.

Re: Clearview AI being used to identify Capitol rioters

#74
post #5

It seems that face recognition could be as dangerous as country laws are. China's social credit system is a strong mirror to its government. The Black Mirror scenario with all those people scared to even talk publicly could (if once real) be strongly country specific, if we get to the point, where life drastically depend on ethics of technology laws.

I think we already have a system in the West where people are afraid to speak openly; it’s referred to as cancel culture but exists elsewhere. The difference is that it’s not a formal government system but more of a mob social justice system.

Is there any reasonably valid speech that is unduly silenced by cancel culture? I don't disagree that people are restrictive of their speech, but I am trying to ascertain if we are better or worse off due to self-censorship.

Re: Clearview AI being used to identify Capitol rioters

#75

Earlier quoted context omitted.

> I want to preserve inequitable outcomes in policing because society is stronger with inequitable outcomes - permits good life for high percentile individuals and constrains operations for low percentile individuals. Thanks for being clear about your perspective. Do you think there's potential for abuse with different rulesets for different people? > Demarcating crime from uncrime is Sorites paradox. I disagree. The…

> Thanks for being clear about your perspective. Do you think there's potential for abuse with different rulesets for different people? Oh, most certainly. The same structure allows for racial discrimination, which I do not believe is a sensible angle of discrimination: i.e. I think Ben Carson should not be discriminated against for being black. Too high value as a top surgeon. On the whole I accept it, though, becau…

> some people are not capable enough to handle the responsibility.

Unfortunately I'm not convinced that anyone is capable of handling the responsibility of determining who is / is not capable of having responsibility. There's just too much potential for abuse.

Thanks for the reply.

Re: Clearview AI being used to identify Capitol rioters

#76

Earlier quoted context omitted.

> Thanks for being clear about your perspective. Do you think there's potential for abuse with different rulesets for different people? Oh, most certainly. The same structure allows for racial discrimination, which I do not believe is a sensible angle of discrimination: i.e. I think Ben Carson should not be discriminated against for being black. Too high value as a top surgeon. On the whole I accept it, though, becau…

> some people are not capable enough to handle the responsibility. Unfortunately I'm not convinced that anyone is capable of handling the responsibility of determining who is / is not capable of having responsibility. There's just too much potential for abuse. Thanks for the reply.

Thank you for the discussion.

Re: Clearview AI being used to identify Capitol rioters

#77
post #4

I realize that this is one of those situations where I'm supposed to be supportive of the use of facial recognition, but I'm still uncomfortable. Facial recognition seems like an all or nothing deal to me, and even after the capitol has been raided... I still opt for 'nothing'.

Is it that different to wanted posters or publishing cctv of a crime? I get why people don't want facial recognition cameras on every street corner but using it to identify people from pics is a different activity... Edit: I completely agree there is nothing special about this case. No exigent circumstances.

I'm more concerned with the data being in the hand of an unaccountable private company than with the facial recognition itself to be honest. I remember that Clearview AI itself was associated with far-right individuals last year, so having them police a far-right insurrection from their personal black box system is more than suspect.

If it happens, it should be done by an institution that is under the supervision of congress and staffed by public servants. This emerging, largely unaccountable surveillance industrial complex with ties to extremist political figures worries me more.

Re: Clearview AI being used to identify Capitol rioters

#79
post #54
post #37

Earlier quoted context omitted.

Out of curiosity, why do you feel like you're supposed to be supportive of the use of facial recognition in this case?

Identifying terrorists is a positive outcome.

I love the back and forth of terminology ("rioter" vs "terrorist" vs "attacker") because I think that's the key to this discussion. The military is given access to dramatically more advanced weapons and technology than police. When the US has suffered such an obviously violent and direct attack (vs planning or "alleged"), is that not a reason to loosen the restraints?

However, it's hard to put the genie back in the bottle with technology. Like we've seen this summer with local police using military hardware, there will always be the urge to inappropriately deploy military-grade technologies in a civilian setting.

Re: Clearview AI being used to identify Capitol rioters

#80

Facial Recognition Tech is obviously good if used by competent people: Narrows the number of haystacks to search for needle => Reduced resources required for successful search => More crimes prosecuted. Problem occurs when: * Users of FRT assume all in haystack are needles * Crimes on book must not be universally prosecuted The first part can't be helped. US Police, like most US government jobs, is a rest-and-take-it…

I really can’t tell if this is sarcasm, satire, or serious. I think satire?
Post reply on HN