"A positive prediction should launch an investigation, not put someone behind bars directly." That is the scariest thing I've read. All we need is a black box to investigate people at anytime.
Yup, this sounds like software driven Minority Report.
Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
21–30 of 54 posts
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#22The number of people killed by terrorism in developed countries per year is, what, a hundredth of those killed in car accidents? A tenth of those shot in Chicago alone? http://www.businessinsider.com/death-risk-statistics-terrori... Terrorists have to kill literally hundreds of times more people per year for it to justify even the meagerest response. And one may argue about the economic impacts, but that's just a ref…
If the media wouldn't cover terrorism it would probably go away by itself. Israel is probably the only country dealing with it effectively.
So if the media weren't flagrant whores for eyeballs things might be better. I'm not holding my breath. The fable of the farmer and the snake comes to mind.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#23The number of people killed by terrorism in developed countries per year is, what, a hundredth of those killed in car accidents? A tenth of those shot in Chicago alone? http://www.businessinsider.com/death-risk-statistics-terrori... Terrorists have to kill literally hundreds of times more people per year for it to justify even the meagerest response. And one may argue about the economic impacts, but that's just a ref…
If the media wouldn't cover terrorism it would probably go away by itself. Israel is probably the only country dealing with it effectively.
So if the media weren't flagrant whores for eyeballs things might be better. I'm not holding my breath. The fable of the farmer and the snake comes to mind.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#24The number of people killed by terrorism in developed countries per year is, what, a hundredth of those killed in car accidents? A tenth of those shot in Chicago alone? http://www.businessinsider.com/death-risk-statistics-terrori... Terrorists have to kill literally hundreds of times more people per year for it to justify even the meagerest response. And one may argue about the economic impacts, but that's just a ref…
If the media wouldn't cover terrorism it would probably go away by itself. Israel is probably the only country dealing with it effectively.
Author here. These are great points and I tend to agree. While the tech is interesting to me as a way to reconcile privacy and security, changing the incentive structure is a more straightforward initial response.
Perhaps this post is most useful as a counter-argument to any "privacy vs safety" arguments that might be used to justify unrestricted access to consumer data.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#25The number of people killed by terrorism in developed countries per year is, what, a hundredth of those killed in car accidents? A tenth of those shot in Chicago alone? http://www.businessinsider.com/death-risk-statistics-terrori... Terrorists have to kill literally hundreds of times more people per year for it to justify even the meagerest response. And one may argue about the economic impacts, but that's just a ref…
If the media wouldn't cover terrorism it would probably go away by itself. Israel is probably the only country dealing with it effectively.
Not just the media. A lot of current U.S. politicians apparently believe their best path to power is endless fear mongering. And it works pretty well.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#26The number of people killed by terrorism in developed countries per year is, what, a hundredth of those killed in car accidents? A tenth of those shot in Chicago alone? http://www.businessinsider.com/death-risk-statistics-terrori... Terrorists have to kill literally hundreds of times more people per year for it to justify even the meagerest response. And one may argue about the economic impacts, but that's just a ref…
Author here. So the techniques in this post aren't actually limited to terrorism, although it' s probably the easiest to talk about. It's broad enough to include homicides (as mentioned in the post), and hundreds of thousands of people are murdered each year because they aren't able to predict their own danger soon enough for law enforcement to be informed (and to intervene).
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#27I like this, but I would argue that predicting crime isn't the core problem that law enforcement has today. The real issue is the incentive structure that disincentivizes crime prevention. That sad truth is that police bear no cost when they fail to prevent crime and, in fact, get more funding and power if crime goes up.
The machine should be able to make economic policy decisions as well.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#28Interesting, I didn't know that homomorphic encryption was already advanced enough to be feasible for actual computation. Years ago I just read about the proof-of-concept systems that worked but were too slow.
I'm still doubtful if the specificity of such pre-crime systems will be high enough that only a negligible number people will be wrongfully investigated. After all, the prevalence of terrorism is extremely low in any population. I guess if you trust law enforcement enough to escalate the investigation slowly and carefully (instead of putting a suspect on a no-fly list immediately) it can work.
Also, with regard to the audits by NGOs or government watchdogs: I suppose you would also need the auditors to cryptographically sign the version of the software they audited so that users can check that a trusted surveillance system was deployed.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#29Interesting, I didn't know that homomorphic encryption was already advanced enough to be feasible for actual computation. Years ago I just read about the proof-of-concept systems that worked but were too slow. I'm still doubtful if the specificity of such pre-crime systems will be high enough that only a negligible number people will be wrongfully investigated. After all, the prevalence of terrorism is extremely low…
Re: cryptographically signing the software version. I think that's a brilliant idea. I'll look into it.
Re: Safe Crime Prediction: Encrypted Deep Learning for Less Intrusive Surveillance
#30For those unfamiliar with homomorphic encryption, I found this to be a good algebraic treatment. https://web.wpi.edu/Pubs/E-project/Available/E-project-04261...