Case Study: Predictive Policing

Loading...

Из курса от партнера University of Michigan

Data Science Ethics

85 оценки

University of Michigan

Data Science Ethics

85 оценки

What are the ethical considerations regarding the privacy and control of consumer information and big data, especially in the aftermath of recent large-scale data breaches? This course provides a framework to analyze these concerns as you examine the ethical and privacy implications of collecting and managing big data. Explore the broader impact of the data science field on modern society and the principles of fairness, accountability and transparency as you gain a deeper understanding of the importance of a shared set of ethical values. You will examine the need for voluntary disclosure when leveraging metadata to inform basic algorithms and/or complex artificial intelligence systems while also learning best practices for responsible data management, understanding the significance of the Fair Information Practices Principles Act and the laws concerning the "right to be forgotten." This course will help you answer questions such as who owns data, how do we value privacy, how to receive informed consent and what it means to be fair. Data scientists and anyone beginning to use or expand their use of data will benefit from this course. No particular previous knowledge needed.

Из урока

Societal Consequences

In Module 8, we consider societal consequences of Data Science that we should be concerned about even if there are no issues with fairness, validity, anonymity, privacy, ownership or human subjects research. These “systemic” concerns are often the hardest to address, yet just as important as other issues discussed before. For example, we consider ossification, or the tendency of algorithmic methods to learn and codify the current state of the world and thereby make it harder to change. Information asymmetry has long been exploited for the advantage of some, to the disadvantage of others. Information technology makes spread of information easier, and hence generally decreases asymmetry. However, Big Data sets and sophisticated analyses increase asymmetry in favor of those with ability to acquire/access.

Societal Impact16:26

Ossification7:24

Surveillance4:54

Case Study: Social Credit Scores7:59

Case Study: Predictive Policing8:49

Познакомьтесь с преподавателями

H.V. Jagadish
Bernard A Galler Collegiate Professor
Electrical Engineering and Computer Science

A major place that data science brings value is in being

able to deploy resources most effectively.

Police work is no exception and predictive policing

has become a very popular term with resource-strapped police

departments trying to deploy their forces in the most effective manner possible.

Now the term predictive policing may suggest that the police are able to

predict precisely where and when and by whom a crime will be committed.

The kind that one might expect from say a pre-crime unit

in the movie Minority Report as we see in this clip.

>> Imagine a world without murder

[MUSIC]

>> I lost my best friend!

>> I lost my aunt!

>> I lost my dad!

>> I lost my father!

>> I lost my wife.

>> Just 6 years ago the homicide rate in this country had reached epidemic

proportions.

It seems that only a miracle could stop the blood shed, but

instead of one miracle we were given three, the precognitives.

Within just one month our pre-crime program,

the murder rate in District of Columbia was reduced 90%.

>> They were going to be waiting for me in a car.

>> He was going to rape me.

>> I was going to be stabbed.

>> Right here.

>> Within a year pre-crime effectively stopped murder in our nation's capital.

>> In the six years we've been conducting our little experiment

there hasn't been a single murder >> And now pre-crime can work for you.

We want to make absolutely certain that every American can

bank on the utter infallibility of this system and

ensure that what keeps us safe will also keep us free.

[MUSIC]

>> Precrime, it works.

>> It works.

>> It works.

>> It works.

>> It works.

>> It works.

>> It works. >> Precrime, it works.

>> On Tuesday, April 22nd, vote yes on the National Pre-crime Initiative.

>> In practice, predictive policing doesn't mean that you're going to have

an accurate and precise prediction of the type that we just saw.

Such predictions take away free will and surely

we all are, even the criminals amongst us, thinking beings.

With choices that we make in life, rather what predictive policing can do for us

is predict probabilistically, where and by whom and

when, certain crimes are more likely to happen than normal.

[MUSIC]

[SOUND] >> Okay Jed,

what's coming?

>> Double homicide, one male, one female.

Killer is male, white forties.

Agatha nailed the time frame at 8:04 AM.

The twins are a little fuzzy on that so we'll need confirmation.

Location still uncertain.

Remote witnesses are plugged in.

This will be case number 1-1-0-8.

[MUSIC]

Morning, detectives.

[MUSIC]

Case number 1-1-0-8.

Pre-visualized by the precogs, recorded on HoloSphere by Pre-Crimes [INAUDIBLE].

My fellow witnesses for case number 1-1-0-8 are Dr.

Kathryn James and Chief Justice Ryan Pollard.

Good morning.

>> Good morning. >> Will the witness preview and

validate number 1-1-0-8 at this time?

>> The private and I will validate.

>> Go get him.

>> Stand by.

>> Time of murder 8:04 AM.

>> Just working with the probabilities is enough

to be able to deploy resources more effectively.

In terms of having police presence in areas where their presence

is most likely to be the most valuable is probably a good thing all the way around.

However, one can go further.

And there are cases where

predictive policing have tried to identify Individuals and

suggest to them not to commit crimes when the algorithm has suggested that

certain individuals might potentially be so inclined.

>> When the precogs declare a victim and

a killer their name is embedded in the grain of wood.

Since each piece is unique the shape and grain is unique.

The shape and grain is impossible to forge.

>> I'm sure you all understand

the legalistic drawback to pre-crime methodology?

>> Here we go again.

>> Look, I'm not with the ACLU on this, Jeff.

But let's not kid ourselves.

We're arresting individuals who've broken no law.

>> But they will.

The commissioner of the crime itself has absolute metaphysics.

The precogs can see the future and they're never wrong.

>> But it's not the future if you stop it.

Isn't that a fundamental paradox?

>> Yes it is.

You're talking about predetermination which happens all the time.

[NOISE] [SOUND] >> Why'd you catch that?

>> Because it was going to fall.

>> You're certain?

>> Yeah.

>> But it didn't fall, you caught it.

The fact that you prevented it from happening doesn't change the fact that it

was going to happen.

>> You ever get any false positives?

Someone intends to kill his boss or his wife but they never go through with it.

How do the precogs tell the difference?

>> Precogs don't see what you intend to do, only what you will do.

>> Then why can't they see rapes or assaults or suicides?

>> Because of the nature of murder.

There is nothing more destructive to the metaphysical fabric that

binds us than the untimely murder of one human being by another.

>> Somehow I don't think that was Walt Wittman.

>> It was Zara Cinnamon.

She designed precogs, designed the system and pioneered the interface.

Speaking of interfacing, I'd love to say hello.

>> To Hineman?

>> To them.

>> The question is not that somebody is being put into prison or

being punished before they commit the crime.

We all do things that sometimes break some minor rule or the other.

Most of us get past these minor infractions

with no consequences, and this is just the way society functions.

The thing with probabilities is that they're probabilities and

just to get a sense of how this might make sense and

where this might cause a problem let's just make up some numbers.

So let's say that the population as a whole,

that one in a million people will commit a murder.

Let's say that some person because of friend

circle and life situation, that algorithm has determined

they are a thousand times more likely than the average person to commit a murder.

A thousand times is a large factor and certainly it makes sense for a police

department to pay more attention to this person than to the average citizen.

Recall that we said it was a one in a million chance for the average person.

And this person for whom the chances are a thousand times average,

it's still 1 in 1,000.

So here you have somebody who is

far more likely to commit a murder than the average citizen but

this person still has only 1 in 1,000 chance of committing a murder.

At this point, treating this person as if they were a murderer or

that they were likely to commit a murder tomorrow.

Just isn't appropriate because the overwhelming odds

are that this person, even though far more likely than the average person to commit

a murder Is still extremely unlikely to commit a murder.

Probabilistic predictions can help and they can help particularly in

the aggregate in terms of deploying resources, including police resources.

When one is trying to exploit the value that one gets from such predictions,

one should though remember what's probabilistic,

and that probabilistic is very different from definite.

And keep that in mind in terms of determining what the consequences to

individuals are.

Ознакомьтесь с нашим каталогом

Присоединяйтесь бесплатно и получайте персонализированные рекомендации, обновления и предложения.

Coursera делает лучшее в мире образование доступным каждому, предлагая онлайн-курсы от ведущих университетов и организаций.

© Coursera Inc., 2019 Все права защищены.

Загрузить из App Store

Загрузить в Google Play