the outreach and the same ausa was prosecuting a muslim individual who used to participate in the outreach efforts. they nabbed him for some immigration information. so it was a huge conflict of interest. but they walk in with straight face and say, what's the problem with that. a huge at least legitimacy problem. and then you have false statement prosecutions. they will use the meetings to go and meet with people the way that law enforcement recruits informants, they'll meet with people in a non-threatening environment, get to know them, build an informant. then they'll give voluntary individuals thinking that they're trying to help the police. they'll make a false statement about something material that isn't necessarily related to terrorism. did you go to yemen? no. turns out he went to yemen to visit his mother, but he was scared to tell them. he thought they'd think something bad. they use that false statement, that's up to five years in jail.

if he cooperates, they'll, if he's willing to be an informant, they'll drop the charges. it's a very dirty game. very high risk for the community. now there's a whole trend of denaturalization. we're starting to see a troubling trend, muslims and arabs and south asians, they've been here for 20, 30 years and go back to see if they can find a false statement and they go after them hard. it's the al capone method. we'll find anything we can. it gives them the in into the community too find the bait. it's very problematic. okay. i have run out of time, so i'm going to skip the civil rights implications because obviously with all the civil liberties from the government infringements it creates civil rights. but i'll talk a little bit about the assumptions. first is, this entire program

assumes that, one, muslim home grown terrorism is a serious problem when the facts and the data show otherwise. the vast majority of those who have attempted to engage in terrorism on u.s. soil are either foreigners who have come are recently here or part of a sting operation. there's a small minority that in fact people born and raised here and they got involved in a terrorist act. that's in addition to the fact that there's more numbers of non-muslim, i hate to use the word home grown terrorism, but terrorism suspects that are actually not related to islam whatsoever. it assumes these terrorists are freely operating and talking within muslim communities as if they're hiding them. when in fact, for example, the youth who often are victims of sting operations who will go online and they'll try to get

them to go to -- to the middle east and it's usually an fbi undercover agent doing that and then met them at the airport, their parents have no idea this is happening. their families don't know. their religious leaders don't know. so to go and ask the community, can you snitch, separate from how problematic that is, it's not even effective. and then as if the community is collectively responsible. so it's again, the collective punishment. okay. sorry. i obviously -- i just want to end with the shared responsibilities committee which is now where the fbi is proposing for muslim communities to put together civilian committees to go and make interventions with youth that are radicalizing or being recruited online. there are absolutely no safe harbors. so what happens if you help a youth that you think is trying

to be recruited by a terrorist group and that youth goes and engages in the terrorist act. you couldn't stop them. could you be prosecuted? what about the victims of the terrorist act? could they prosecute you civilly? what guarantees do you have that cooperation with the fbi in good faith doesn't end up putting you in jail? and do you think that the american society will stand up for your rights if they find out that the fbi betrayed your tres? they will not. they will not. so this is the apparatus. and it's very, very entrenched. so i would just end by saying this is something that you-all should get to know more about because it's coming to a neighborhood near you. and, you know, first they come for them. then they come for us. and then we all lose. thank you. [ applause ] >> please join us in a round of

applause. thank you. [ applause ] thank you, professor. i just want to say as we tran to our next panel, i think one of the really important things to reiterate is something i think people in this room know but it bears repeating. when you hear muslim, we need to hear a very large number of african-americans. and so there is a significant overlap here. and my understanding -- i'm sure folks in the room will tweet out the right statistic, anywhere from a fourth muslims are african-american. i am tremendously excited about this next panel. as i mention in the morning, a lot of philosophers, theorists, writers have done a lot of thinking about surveillance and about watching. and yet so few of this them have realized the racialized nature of that watching. i think it's -- it's always risky to talk about 1984, but i will for this purpose.

the curious thing -- wif the many curious things about 1984 is that the area of that society -- what's the name of it? anyways, the name of that society, the least surveilled area is the area where the working class lives. i think that is one of the unfortunate aspects of that betrayal. we are joined by professor browne who has done powerful work in updating that old surveillance to account for that racial gaze and we're joined by professor obasogie. he asked a very interesting group to figure out how we see race, blind people. and chairing this is my co-host,

professor paul butler. a round of -- >> thank you. [ applause ] >> thank you, alvaro. now time for another deep dive. we've heard from historians, activists, scientists and technicians. we wanted to be a little creative and add some theory to the mix. we have two amazing scholars doing some of the best work on the subject of what we talk about when we talk about surveillance. i guess i've always thought of surveillance as a visual technology. i think a lot of us did until osagie had this brilliant idea. he asked a simple question that nobody had asked before, how do the blind see race.

in his book "blinded by sight" disrupts the way that we understand race and the way that we understand what it means to see. so he's going to tell us about what he's discovered. >> well, thank you, paul. it's really good to be here and i'm really excited to be part of this panel. in the aftermath of pearl harbor and the united states formally entering world war ii. [ inaudible ] -- but this created a bit of a problem for many americans. how can we tell the difference between chinese people who are our allies and japanese people who are now our enemy. from an american perspective two indistinct groups had to be separated. their bodies and movements had to be watched closely as a matter of national security. so life magazine stepped in to

assist in this project of racial surveillance. life published a multi-page spread on how to scrutinize asian bodies to tell the difference between friend and enemy. you see various kind of detailed markings on a chinese person's face to tell the difference, similarly with the bodily depictions. so this example highlights the deep and complicated relationship between race and surveillance. moreover, it shows the political nature of seeing race. that is, how racial bodies become seen as visually obvious because political circumstances change the way that we see. seeing and surveillance are often understood as mutual or natural engagements with the world around us. it is thought they are simply observing and collecting information about activities that are visually obvious. as this example from world war ii highlights, seeing an observation may very well have

little to do with visual perception. our ability to see race and the attention we paid to it may come from something other than the notion that race is something that visually obvious. to explore this idea, i conducted a series of interviews with people who have been totally blind since birth and asked them about their understanding of race and racial experiences. this research is the first time that anyone explored the way that race is understood in the blind community. it is largely assumed that race much diminished significance to blind people. this assumption is the genesis

of the popular color blind probe that we see in law and public policy where being blind to race is thought to bring a racial utopia. we think of blind people as folks that people that treat people according to character and not their skin color. they may be able to speak to the social practices that inform their understanding of race. since these are practices that similarly affect sighted people, but are less accessible to them, blind people's experiences can sharpen our understanding of how race becomes visual for all people. moreover in relation to this conference, this can provide insight into the social conditions that make surveillance of racial bodies possible. so a quick note on methods. so the target population for this research were blind people who have been totally blind since birth. i also interviewed a small sample of sighted individuals to examine whether or not blind people's understandings of race are different from sighted people. i interviewed over 100 blind adults of various ages, backgrounds and residences. both were selected through snowball sampling. each interview was recorded with

the respondent's consent. okay. so the first step in this project was to establish two common sense approaches to race within the sighted community. first, that sighted people have a visual understanding of race and second that sighted people think that blind people have a diminished understanding of race. these were established through my interviews. in the interest of time, it is clear that sighted people think that race is visually obvious. let's move onto the core research question. race is understood and experienced by blind people the same way experienced by those that are sighted. that is visually. they associated race with skin color and other visual cues. many of you may be thinking that any visual understanding of race that blind people might have reflects a general awareness of how the rest of the world

operates. blind people's visual understanding of race profoundly shapes the way they think about it. it also affects their response to race at deeply emotional levels. the blind respondents i spoke with gave similar answers. one person said race is a way of dividing up human beings according to color of skin. another said race is skin color, color of one's pigmentation. another said race is color. even though i can't see it, that's what i tend to think of. most blind respondents went in visual terms. to have a more sophisticated understanding of the range of visual cues. one blind respondent defined race as physical attributes that make people different from each other, skin color, maybe some of the physical features that make people different from one another. another noted, it's not only skin color because it's also other characteristics. i know the black race has facial structure and body structure. i know that each race has its set of characteristics to go with it.

race is not only based on color. these passages highlight how blind people often have a nuanced understand -- to signify race often as sophisticated as their sighted peers. what becomes apparent is that the ability to see the markings that define racial boundaries is neither necessary or sufficient in explaining the strong association of race with visual cues. if blind people define and react to race in visual terms, then the empirical evidence pokes holes in the assumption that race is visually obvious or self-evident. other sensory experiences also affirm the importance of race

for blind people without displacing the visual significance. while no sighted respondents identified voice, over half of the blind responds did. this should not be surprising. what is surprising however is that these audible clues do not stand in for the visual cues nor do they become primary in how blind respondents conceive race. voice and accent remain secondary measures. so one respondent said voice and accent doesn't mean anything to me except that i know they have different skin color. another said, as i got older, i realized or learned that voice is not a very good way to identify swin because it's not reliable.

okay. so blind and sighted people are part of the same social fabric that directs individuals to pay inordinate attention to visual cues that signify a racial difference. this process is effortlessly transparent for sight the people and makes the social experience visually obvious. but it takes a bit more work for blind people, excuse me, reducing race to visual cues. as a result blind respondents are capable of describing the social practices that give the visual cues associated with race a feeling of obviousness. one said, quote, that's what people talked about when i was little and first introduced to people of races other than my own, they used terms that had to do with skin color. it became part of the respondent's racial vocabulary. it also shapes the underlying meanings given to racial labels that informs a visual sensibility given to blind people. another blind respondent notes, i was brought up to learn i was

white of course. i learned that i was white so that white could be contrasted with black. i was driving with my father downtown and he said do you smell that smell. he said that's the smell of nigger town. i didn't know what that meant but he was perfectly glad to tell me. that is why the negro lived. then he began to describe all the stereotypes of being a negro. at that time there was a difference. if you were pretty good, you were a negro, otherwise you were a nigger. it didn't matter, you still weren't a white person and that's the way it was. he would say, you know, what you smell is partly the way they keep their houses and yards and there's trash laying all around. then part of what you smell is just them, they can't help it. he would go on, they talk differently because they're less educated, less capable of being educated.

pretty soon you begin to develop a race identity that's wow, this is sad for them and sad for us too. visual differences can become vividly real even for those that cannot see. it informs differences that cannot be seen so that they feel like common sense. a belief system that race is visually obvious is being structured. the experience related by blind respondents are not unique to the blind community. rather, they reveal how all individuals are trained to seek and give meaning to the visual distinctions that society deems important. win respondent provided an example, quote, we had a babysitter -- excuse me. we had a babysitter named ellen who was black. i came down and said to my mother, what are you doing. she said, i'm washing the counters. she said, black people smell and your babysitter was here last night.

i said, that's interesting and filed that a way. ellen came back the next week and i walked up to the counter and sniffed it. she said, what are you doing, i said i'm sniffing the counter because my mom said you guys smell and she was right. this illustrates how differences didn't make a difference until it was pointed out and racialized. it does not take a fantastic leap of logic to see how these social practices create a visual sense of racial difference among blind people and make visual cues seem obvious boundaries. in many instances, race was socialized to not only take on a visual significance, but a deeply emotional one that impacts their everyday lives. one area that's particularly revealing is in dating and romantic relationships. one blind black respondent said, quote, i just love african-american women. i don't know why. i had white friends that i hung out with and we went to class together and worked on projects together.

i just never had a desire to do that. i think i did it about a week and i was like, no, i can't do this. this respondent explains some of the cultural barriers that make interracial dating difficult. cultural barriers can be difficult to transcend. by this interview revealed a difficulty with race as it plays out in physical differences, not merely cultural ones. other respondents voiced this hesitation as a desire to not disrupt social norms. it is nonetheless looked down upon in society. one respondent recalled a white friend's experience, quote, he was going to college and started working with a reader. she was very attracted to him and he started seeing her. then somebody told him she was black and he broke it off. he justified it by saying it

would not have worked in the south. where a white plan could be involved with a black woman. once he learned she was black, prejudice set in. how race becomes a primary filter for dating within the blind community. he said, quote, a lot of my black blind friends have sort of a joke because when someone doesn't know our race, especially males they'll find out a way to reach out and touch our hair. i go to conventions now, national conventions for blind and people trying to meet somebody to date. you can see they're pursuing somebody they find attractive and go for their hair and then change their mind. they're still friendly. i never known anyone who stopped talking to anybody altogether. they'll give themselves some time, but, you're black. what does this mean? this project's empirical contribution can be summarized by one of the blind respondents

who said, quote, race is not a mystery to blind people, was kind of sad. sometimes sighted people look at blind folks and think these team people can show us to a star trek race blind society. it be great if we can do that. but we're just as much a victim of racial prejudice, stereotypes and misconceptions as anyone else. the fact we're not clued into it directly by vision doesn't change that a bit. another blind respondent, quote, race plays just as important part for blind people as it does for others. i wish we could be the societal model that would show every society who gives a damn how to be color-blind but i don't think we can. there's more to race than what's visually observed. we built race whether it's there or not. this highlights the project's key finding the presumption that race is obvious is part of a social process that produces a visual understanding of race, at the same time it masks its own existence by making race seem obvious.

these perceived visual distinctions are social practices that are so strong that even blind people see on organize lies around visualized race. rather than being obvious, seeing race is social rather than visual phenomenon. the salients of race is linked to social practices that produce visual understandings for race in blind and sighted people. i talk about this "blinded through sight, seeing race through the eyes of the blind," finding provide a basis from which to start questioning a key premise of surveillance policies and technologies. practices work from suggestion that visual surveillance is in a sense merely observation or neutral assessment of people's behaviors. this work shows how seeing and vision are inherently political processes constituted by social norms. surveillance of marginalized people entails political process

of creating visible black and brown body that becomes a target. resent on race and blindness shows how visibility of black and brown criminal bodies is produced by broad political narratives rather than anything obviously seen in the community. by engaging expanding forms of surveillance, the state creates suspicious behavior it says it observes. this is why surveillance must be questioned and resisted as a project of racial jut. thank you. i look forward to your comments. >> thank you. >> thank you so much, osagie. that was fabulous. look, a negro, that famous line took simone brown on a journey that she details in her book "dark matters."

how could you not love a book that has a chapter called "everybody got a little light under the sun?" everybody's got a little light under the sun, a famous song. another within called "what did the tsa find in solange's throat"? simone, what did the tsa find -- [ laughter ] >> thank you. i don't have an answer for that one, but thanks for sharing it. i'm going to talk, briefly, about a couple of moments to get us to a jumping off point to talk about how people often critique surveillance often using very tools of surveillance. some people call that a surveillance, inversion of that. two instances. study came out on twitter yesterday, they -- a psychologist department at uva interviewed medical students to ask them these questions around if they think that blacks age more slowly, do whites have

larger brains than blacks, are black's skins thicker than white. stereotypes, the idea of the super predator, these things i want to link it to another story that came out as well there is week. this is an image of a recording and the transcript of eatle easter, she had gone to august, complained about issues around hernia and she had gone to see a surgeon and she was really kind of disappointed the way that he talked to her. when she went to have the surgery two weeks later, what she did, so it gets back to solange, she hid within her

braids a recorder about the size of a usb recording what they were saying while she was on the operating table. you see use of surveillance technologies to record. what came out of moments were quite shocking for her. they spoke about her being a queen, called her pejoratively precious, in reference to the film, made reference to bill cosby, innuendos of unwanted touching and taking photographs. these are moments when she used technology to kind of turn a gaze on the medical model. we talked about the racial gaze earlier on, paul, you mentioned it. you can see how the gaze takes on a particular medical dimension. look at one of my colleagues, sarah brain, using longitudinal data, the ways if which they avoided places like hospitals, like other institutions, and you can see kind of the link between that matrix of policing and also hospitals as well, too. both of us use this image here, i use it as a jumping off point. this is desi crier, he was a worker in a camping store at texas who tested out this technology, h ptechnology, for facial automation. he was trying to see if the camera would zoom, tilt, and pan the way it was technically supposed to. but he found it wouldn't follow him.

when his colleague, so he called himself black desi and colleague white wanda. when she would enter the frame, the camera worked properly, able to adjust to her movements. it was unable to read dessy. i look at these moments, i think what happens to surveillance when we question conditions of blackness. when we look at how blackness enters the frame. some there are other moments in which you have this prototypical whiteness. i know the next series of questions will look at biometric technologies. the idea that this is also about automation as well, too.

i'm going to skip ahead i want us to have a conversation and i hope the other group will take up moments of the memiification of prototypical whiteness. a moment last sum, an intersection here, intersection of river rock trail and dunes drive in the craig ranch development of mckinney, texas. 15-year-old was thrown, kneeled on and detained by a mckinney police officer. many saw the video to detain a girl in a bikini at the time. the original clip shot by a 15-year-old boy, brandon brooks. he says everyone who they were putting on the ground was black, mexican, arabic. the cop didn't even look at me. it was kind of like i was invisible. i'll close there. thank you. >> so, get us started in the conversation and quickly invite your comments.

please, think of questions for this amazing group. we thought a lot about blackness and the white gaze. i'm curious about how the white gaze relates to white people. so some critical theorists said we don't define whiteness other than it not being of color. so we have lots of constructs of what it means to be black. not so many constructs what it means to be white. so i'm curious about whether the blind see whiteness, how they think about that. >> interesting question. so the blind white respondents i spoke to exhibited what was termed racial transparency in her piece from 1990s she talks about how whites are unable or unwilling to see race or their own race or whiteness as a race

but see other people as racialized as black or latino. the same phenomenal i saw in the blind and white respondents i spoke to, they were able to talk about race as something other communities have but didn't see themselves as a race or whiteness as a racialized experience. so, that's again another example of the kind of parallel racial experiences between sighted and blind people and how that stems from a similar social experience in terms how whites are racialized and socialized and think about race something other people have. >> you both have an idea that seeing is political. it not biological. it's not impure cal. it's not empirical. it's value laden. so, when we think about what we

want to happen on the ground, the kinds of transformation that we need, what's the corrective? is there a need to be a corrective? are there glass that could help us see better? >> great question. one of the things that i try to do my work is to have a shorthand that we can talk about and i think technologies that -- this is why you asked me to move that -- yes. thank you. i looked at so many images i flew by quickly were biometric technologies. what changes we can have a critical understanding of how they're put to use, ask questions how they are stored, shared, sold. understand technology of something. when people say algorithm we understand what it is that they're talking about. we can question things like

those moments. one of the things i went through was this last summer, somebody using one of google's apps upload a photo and automatically tag that photo, a building or bicycle and this is a black man. when he would upload pictures of his friends it would tag them as gorillas he under stand what was happen, training data that -- what kind of training data the technology being fed into this program to understand to read certain black faces as gorillas and there are long histories of doing that as we know. he called them out on twitter using the same technologies to question what was happening there. it's happening that space to call out and offer corrective for ways in which technologies seem to be designed or by design to put certain bodies around gender, race, able bodiness, they're not outside our understanding 0 of race. >> one or two questions from the

audience? yes, sir. >> yeah, thank you very much for talking about it. but i think, as we were doing that, i hope this doesn't get lost being a white and black issue and the whole racial gaze as well because it's also something that while within the context of white supremacy all of the people of color are less than human, but i think we also need to separate what does people of color mean. i think it whitewashes a lot of hierarchy that is within the people of color community as well. i mean, how muslims in south asians and arabs look at black people. how latinos look at black people. how asians look at plaque people

as well. i'm hoping when we talk about surveillance and racial gaze, it very much is apart of than earlier somebody said the muslims are in the new black. no, they're not. it's like saying all lives matter. absolutely not. then at least like just fight back the system as blacks have been doing. so i want to throw that in the mix we don't want to get lost in this thing and glaze over how other people of color look at black folks and what happens with that. thank you. >> time for one more question. anyone have a question. >> hello. i guess my question's related actually but it was in working with blind people and talking to them about race, i was wonderinging if you had any findings where they expressed like different some understanding of the way people even within a given racial

category are treat differently because of how say dark their skin is or what features they have or the way their hair is or how much they conform to the like societies arc typical idea of what a black person looks like or a white person looks like. is that level of nuance also there in their understanding or more like black and white as it were? >> yes, there was bit of nuance. themes such as hair texture or skin, how skin feels in terms of roughness versus other, that was certainly there. it was one of those things present but was much more prevalent in the conversation where these kind of broad measuring sticks of who's black, who's white, who's latino, who falls in the various categories. how human diversities can be put in the categorizations. >> actually, i'm informed we have a couple more minutes. if people have other questions, while you're thinking of questions, when i was listen to osagie talk, i was remembering a familiar question from my mom when i was a kid growing up in the old days when there were land lines and no caller i.d.

someone would call, want to speak to my mom. i'd say, ma, someone's on the phone. she'd said, black or white? black or white? and that was significant. the answer -- it gave her information that she needed. and holding that kind of coding when people of color do it, it's sometimes a survival skill. >> hi. i just wanted to piggy back off of this gentleman's question. as far as when you, i guess, conducted your research on with the blind people, did you find that some of them even had

thoughts of other than you know texture of hair and skin, did they have other types of stereotypes, i guess, what -- how certain people are, even though they have never seen, you know, these people? to me, it's at least i work with a blind person i find some of the comments she makes she couldn't have known if someone didn't tell her this or it wasn't engrained in the way that she was brought up. did you find some people said things and you were like, how would you know that if you can't see? >> right. that's an interesting question about the social construction of race, how do certain types of social meanings attach to certain kinds of body. that was prevalent in the community. that's a conversation how we're socialized to react to a certain

type of racial bodies and that's something that happens outside a visual process. it's a learned practice and it's something, again, sighted individuals and blind individuals being part of the society, that's what we're getting. what my research gets at talking to blind people their ability to understand races of visual phenomenal, that's the constituentive understanding of race. the question is how certain bodies become seen as visually salient to begin with. my resent shows even process of coming to see racialized bodies as distinct, different, salient, that is a deeply social and political process, so much so blind people come to see race differently. they talk about race in visual terms though it's not something they can directly perceive. the work i've done shows how this is a parallel process. since blind people don't have access to vision, they're more readily able to talk about the social and political experiences that produce that visual understanding of race but this is the same experience that we all go through. >> one of the things that you raise with you interview with the blind, the slide with the interviews of ople who are doctors or residents in the ways in which they thought about black people not being able to

feel as much pain or healing quicker top biggie back off of that question, the effects of that for sighted doctors people receive less painkiller medication, improper care. those are the outcomes, the material outcomes of these processes how blackness is understood in the hospital. >> what were the numbers representative of on the slide? >> let me bring it back up. >> percentages? >> while simone is accessing the slide, another question? >> i'll give you a rate, it was to bring out the questions that i wanted to, framing of the questions. i don't have the -- all of the data of how many people that they were -- the end number. >> hi. my question goes back to a lot of different things, especially the recidivism risk, the idea if

it's model, it's scientific. and objective in some weird way. it goes back to whiteness and experience of being white, i'm sure you're familiar with la tanya sweeney's research and googled her name and the google ads were criminal record and she googled white sounding names and there were no such ads. part of the response of google is there is no response of google. there is no responsibility though we might call this a racist algorithm there's a lack of responsibility. i feel like that is directly tied and tell me if you agree to the fact that the people who work at google are almost entirely white. >> there's a woman allison bland or ali bland on twitter from princeton. she has a tweet, she knew there were no black engineers at google when it said right turn onalcolm x boulevard. critiques of satire.

>> on that note, please tell me thank our panelists. >> that was fantastic. thank you so much. so, in the morning, i mentioned that i think one of the under explored -- not unexplored -- is the use of surveillance technology originally developed for national security surveillance for military surveillance in the domestic context. i want to take a running tally here. we've heard about two of the technologies. we've heard about predictive policing technology, originally developed to detect hot spots in the battlefields of afghanistan in iraq. and is now used to detect hot spots in the inner city. we learned about stingray technology, who thanks to freddie martinez of lucy parsons labs, which you should all

follow now, is using free dole of information act to finger out how stingrays is being used on the streets of chicago in predominantly low income black and latino communities. now hear from my wonderful colleagues from the center on privacy and technology, claire garvey and jonathan frankel on a third technology that is being used in this manner, facial recognition technology. >> thanks. alvero. so i want to set up this discussion about facial recognition technology which a bit of a hypothetical. so imagine you're walking down the street in a town or a city, something we do all day, or daily, heading home from work maybe, going to a doctor's appointment, maybe attending a political rally. generally speaking when we engage in this activity we do so with the assumption that we're doing so anonymously.

we have relative anonymity. sure, we are presenting our face in public and may come across a co-worker or neighbor, we'll say hi, they'll identify us but we don't think that we're going to be singled out, identified, and certainly don't expect to be tracked by just preventing our face in public. now think about all the cameras you're probably passing by especially in larger cities. these are traffic cameras, personal security cameras, these are police cameras. and imagine that they are zooming in on your face, they're extracting a template of that face and using that template to figure out who you are in seconds. this is being done for law enforcement purposes. so thanks to the vast improvements in facial recognition technology, and in the use of facial recognition technology by federal, state,

and local law enforcement agencies, this is no longer really a hypothetical. this is a reality. for example, this is already deployed in l.a. by the police where a couple of years ago they set up 16 cameras that could use facial recognition technology to surveil in real-time in northern l.a. capable of extracting face template up to 600 feet away. this is not just limited to l.a. these are other police departments be chicago, for example, dallas, west virginia, others have acquired or actively considering choiring this exact capability. this is one type of facial recognition technology. there are others. mobile units that allow for field identification by police officers and then far more common desktop facial recognition systems where an officer can upload a facebook photo, a camera still from cell phone or cctv camera, and these

are very common throughout the u.s. now. so we're here to discuss how this technology runs the risk of disproportionately affecting african-american citizens. the center on privacy and technology now is conducting a widespread research project on how police departments, particularly at state and local level, are using facial recognition. what types of systems they're deploying and then in particular, what policies, if any, they have in place to constrain or to inform this use. alongside this we're also examining the biases that exist in facial recognition technology and risk that the deployment of facial recognition by state and local law enforcement agencies will disproportionately affect african-american communities. so the study's based on a records request that we sent to more than 100 different agencies and we're in the process of reviewing over 10,000 pages of documents.

we're in the preliminary stages but i want to share two initial findings that we have. the first is that there's very wide variation, how the systems are deployed at the state and local level. as i mentioned, there are mobile systems, desktop systems, there are cctv surveillance systems. another difference we find, some of these systems are run against just mugshot databases, others run against entire driver license databases. if you have a license from that state, you're enrolled in a facial recognition database used by state and local law enforcement. anyone here have have a driver's license from ohio, florida, maryland, maybe? yeah. okay. you're in a facial recognition database used by police. the second general finding that we have thus far, and did i mention we're in the initial stages of reviewing these documents? the other general finding that

we are discovering is that there are very few constraints on how these systems can be used. we have yet to see an agency that requires a warrant be issued before they do this type of search. we have yet to see an agency that sets a standard severity of crime before they use this type of system such as wiretap act defense or felony. and then, we have a fair number of agencies that actually have no policy on the book whatsoever, even though they do have access to facial recognition system. they conduct no audits of how they use the system. they don't even keep logs. what we're finding is with some of the agencies not only do we have that much transparency as citizens, as researchers into how they use the systems the law enforcement agencies themselves don't have that much knowledge about how they're using them. so, i'm going to turn it over to jonathan to speak more how

directly this ties into today's conversation about surveillance of the african-american community. >> so we've been taking a look recently about the connections between facial recognition research and today's theme, and i think there are a couple of different aspects i want to dig into. so, the first is just the risk of discriminatory surveillance, disparate impact in the way these systems are used and we have heard almost ad nauseam today african-americans receive disproportionate attention from the police and criminal justice system. one example, pinellas county, florida, one of the longest standing facial recognition programs in the country. african-americans make up 8% of the population, but 25% of the incarcerated population, and what database does the facial system query? the mugshot database which is disproportionately african-american. here's a neat slide claire found from the agency that runs san diego's facial recognition

system. the other issue is this technology's likely to be disproportionately employed on groups already subject overpolicing. people of color 1.5 to 2.5 times more likely targeted than expected presence in the population. to be targeted than expected by the presence in the population so i think that tells you a substantial part of the story right there. now, personally as a computer scientist, i was interested in asking whether the technology itself could also be discriminatory and two of my anecdotes got stolen by the previous presentation but we heard about the hp webcam that couldn't recognize a african-american facial recognition and also the google labelling of african-americans as gorilla as. and these are all very interesting anecdotes and consumer reports followed up on the webcam examples and couldn't validate the results. so there is questions on those -- those situations. but all three of the situations are a result of facial recognition algorithms, the same technology that law enforcement is using. so i was interested in taking a

look at technical research to see if anybody had studied this issue in a rigorous scientistic matter to see if there were issues in the technology. so the first study that i found was -- and the context is every four years nift reaches algorithms and they separated them into two groups, those developed in east asia and western europe and the united states and they wanted to see how accurate the algorithms were on pictures of african-americans and caulk asians. you could see the black lie, that represents the east asia algorithms and more developed than western europe and the united states and the graph on the right, the opposite was true, on pictures of caulk

asians, it was better than that developed in east asia countries. we call this the other race effect. where people are better at distinguishing members of their own race. and facial recognition algorithms have a similar other race effect. and this means it is a setting in which a algorithms is developed could affect its biases and knowing that the mesh software -- american software looks like me, caulk asian and male and this technology is more likely to be employed on african-americans and over-policed demographics. so we could look at another study which asked the question directly where they studied three algorithms that are made by companies that sell to law enforcement and tested on mug shots from pinnelas county florida, in all three gaps, look at the green line and the red line representing performance on african-americans.

and i wouldn't pay too much attention to the blue line because there wasn't a lot of latino represents so there is a difference on graph to graph. you could see a consistent gap between the green line and the red line which raises concerns given that the technology is focused on perhaps the communities on which it performs worst. so what could we conclude from this information? so as a scientist, i want to caution this isn't enough information to convict or indict but enough information to suggest there are technical questions that ask about the way the algorithms work. and another question is how pronounced this effect is. and i've shown you studies and we were scraping the bottom of the barrel so it is hard to say. and are the algorithms being tested for bias. we've had a hard time finding studies and we interviewed companies about the issue and neither of the companies could point to a specific test for racial bias. and finally where does the bias

come from? you could guess the demographic for the engineers, if they are testing on themselves i'm skeptical, the complication used for the training and some people are harder to recognize which certain studies have speculated and one company said this make be the color contrast in the face being reduced for people with darker skin making it harder to distinguish features. so i'll hand it back over to clair to talk more about the study. >> to sum up, facial recognition is beginning to challenge our assumptions or expectations of anonymity in public spaces. and while it has positive implications, it is a powerful policing tool and critical to understand the risks that it poses, particularly if it has this risk, which we're finding in our research to disproportionately impact, particularly african-american citizens. so we'll be publishing a broad report on this.

hopefully this summer, which will include recommendations to the federal government and state and local agencies, police departments, and companies and advocates on how to begin addressing this issue. thanks so much. and i guess we have some time for questions as well. [ applause ] >> your last slide that you had -- does it exist for hispanic as well. thinking about the complex of the issues was on the backs of undocumented migrants and the data base, the secure building the next identification data base to i want to invite you to the extent you could draw conclusions with respect to latinos in the efficacy of

facial recognition generally and not just the extent to which it has been accurate but to what expect migrants and in undocumented communities might be targeted similar to the disproportionate similarity to african-americans. >> and i'll take half of the question and turn to over to clair for the other half of the question. this graph is clean but others were all over the place for the latino study so i wouldn't draw too many conclusions from this study. most of the studies showed it between asian americans and cack asians and. and studies between caulk asians and african-american, this is was the one. there may have been one over pseudo study. with latinos, this is it. there is nothing else i could find in the literature. >> my addition would be an

unsatisfy answer is that we are focused on the state and local level and independent of the ngi system. there is research through eff and epic which i would probably point you to. but this is probably a limitation of the study more than anything else. >> could you talk for a second about what you -- what you mean when you say that the algorithms fails and what the ramifications are of that. i could imagine that actually this might be a surprising benefit if the over-policed communities are actually not being subject to as accurate tracking identification than maybe we should let it slide as long as possible. >> certainly. and that is a fantastic question. so there are two kind of failures they distinguish in the literatu literature. false accepts, where these two people are the same when they are not the same. so it might say this person is the same as a suspect when it is not and false rejects where it says, okay, this person should be the same, but they are

actually not. and this particular study, typically the way they are done, if you want to get technical, look at x axis and see they fix the false separate and show the true separate or how many times they get it right so the way the graphs are framed, you want contrast the false separate to see how often a african-american is misidentified as the correct person when they are not. and if you turn the graph on the side, you could see the false separate is higher but typically the way the studies are structured and we don't have south caroli access to the specific data, it is hard to draw that conclusion. you could speculate this is better if it has a harder time matching african-americans in general but i think that would be a -- i think that would be an overly simplistic way to take a look at the study. >> and to build on that, we could hypothesize that depending on the purpose of the system itself there are different ways to set up the algorithms in the system so maybe sometimes you

want to be overinclusive and underinclusive and i would hazard a guess -- this is a guess, if it is an investigative tool, you want to have more false accepts than rejects because it is more helpful as a tool. if it is not a good tool, they won't use it. >> the other tiny thingly throw in, because we're out of time, the way law enforcement use the algorithms is they measure the similarities between pairs of photos and set a threshold and consider it a possible match. so the problem with perhaps if it is really good at false rejecting for african-americans is that one -- the particular african-american who is being targeted in the search might fall below the threshold but that means that other people who are higher up in the rankings could get implicated instead. so false rejects could turn into false accepts depending on how the system is configured with the similarity threshold. >> let me point your attention to an essay in the atlantic that

ran yesterday with recommendations on how to solve this problem. we're running behind to we'll take a 10-minute coffee break and join us back here a minute after 3:35, so 3:36 and you will not want to miss professor chris hennings presentation. 3:36. see you soon. >> it is known as one of washington's premier events bringing together government officials and members of the press and hollywood stars. live coverage of the 2016 white house correspondent dinner this saturday at 6:00 p.m. eastern, the live coverage from the washington hilton includes red carpet arrivals and background on the dinner and war presentations. 2700 people are expected to attend the sold-out dinner. this year the president will give his final speech as commander-in-chief. join us to watch the 2016 white house correspondents' dinner beginning saturday at 6:30 p.m.

eastern on c-span. >> tonight a hearing on u.s. policy in the asia-pacific region and discussion on new fcc rules for consumer data and later a look at state of literary magazines in today's political climate. deputy secretary of state antoni blinken testified about u.s. interest in the asia patie pacific region at a house foreign committee. they asked about aggression in the south china sea and sanctions against north korea and human rights issues in vietnam and burma. this is just under two hours.