Statistics

AI Getting Too Good At Predicting Crime, So It’s Being Purged

If all we know about an American male is that he is an American male, we can use data on observed rates American males committed crimes to predict that this new American male will commit a crime.

This is not controversial.

Now suppose we have two American men, one black, one white. We can again use data on observed rates of crimes to predict these two men will commit crimes. If we divide the black man’s probability (conditional on this evidence and assumptions) by the white man’s, we’ll come to a number about 10.

This is controversial.

We have just built an AI system to predict crime rates. This is AI, even though it’s very simple. AI, you will recall, is nothing but probability modeling, but given a much more marketable name.

Under the theory of Equality, all men are equal, and when they are not, they are not is because of some cruelty, such as racism, was imposed on the lesser man. It is never the case that the lesser man is responsible for his own failures or shortcomings.

Our AI has revealed an inequality, and thus under Equality, our algorithm is racist. If you find this asinine, you are not an expert.

Experts—that noble breed which rule over us—say that this is so. AI must stop its systemic racism! Yes. Not only that, this: “AI experts say research into algorithms that claim to predict criminality must end“.

A coalition of AI researchers, data scientists, and sociologists has called on the academic world to stop publishing studies that claim to predict an individual’s criminality using algorithms trained on data like facial scans and criminal statistics.

Such work is not only scientifically illiterate, says the Coalition for Critical Technology, but perpetuates a cycle of prejudice against Black people and people of color. Numerous studies show the justice system treats these groups more harshly than white people, so any software trained on this data simply amplifies and entrenches societal bias and racism.

ALGORITHMS TRAINED ON RACIST DATA PRODUCE RACIST RESULTS

“Let’s be clear: there is no way to develop a system that can predict or identify ‘criminality’ that is not racially biased — because the category of ‘criminality’ itself is racially biased,” write the group. “Research of this nature — and its accompanying claims to accuracy — rest on the assumption that data regarding criminal arrest and conviction can serve as reliable, neutral indicators of underlying criminal activity. Yet these records are far from neutral.”

May I translate this for you, dear reader?

AI, i.e. probability crime models, are too accurate when using race. This scourge of accuracy is not science because accuracy is racist. And it must stop.

Springer is about to learn this lesson the hard way. Almost two thousand “experts” wet their pants and sent their soiled underwear to the publisher when the “experts” learned that Springer was about to publish the peer-reviewed and approved work “A Deep Neural Network Model to Predict Criminality Using Image Processing”.

Since it’s often easy for us, and for probability AI models, to tell them difference between black and white, and other races, based on face, the algorithms, and we, too, are racist.

The open letter is the whiniest thing you will see today—we can’t even say “this week”, given the frequency of similar events.

The letter does do the service of proving science is now defined as politically acceptable outcome. What is science today will not therefore be science tomorrow, because what is acceptable today won’t be tomorrow.

“Crime prediction technology is not simply a tool—it can never be divorced from the political context of its use.” “Power is here defined as the broader social, economic, and geopolitical conditions of any given technology’s development and use.” “We borrow verbiage from set theory here to illustrate the deep complexity of such contexts, and to illustrate the peril of attempting to discretize this space.”

My favorite: “One need not harbor any racial animus to exercise racism in this and so many other contexts…” You’re a racist if we say so. We know you’re a racist by looking in your eyes. As long as those eyes are embedded in a White face (do we now also capitalize White?).

This conclusion that you are a racist even when you are not is also the result of an AI algorithm, but in the approved direction. Therefore it is science.

It’s worth a few minutes scrolling through the list of signatories.

So long, Science! You had a good run. But you ran into the Current Year.

To support this site and its wholly independent host using credit card or PayPal (in any amount) click here

Categories: Statistics

37 replies »

  1. Yikes, this (almost) makes you wish for a return of the obnoxious “Question everything!” positivist scientists of the yesteryear. Say what you will about the vile illogical positivist, but at least he accepts empirical data and if empirical data says this group has lower IQ on average, or is on average more prone to criminality then darn it all he will accept it.

  2. 1 Timothy 5: But if any man have not care of his own, and especially of those of his house, he hath denied the faith, and is worse than an infidel

  3. It is possible to have probability crime models WITHOUT race, but it’s never going to happen because those models would address REAL causes that no one wants to speak about. Race is a lazy, political way of dealing (apparently a lazy statistical choice also) with the world and playing into that thought process is not good. I hope Briggs is not advocating the “blacks are awful” movement I see in many people, but rather clumsily trying to address a problem that isn’t real.

    Interesting with the image processing. What if all the images were changed to the same color? What if all the subjects were black, or all were white? The algorithms would fail, would they not? We can predict crime without color, but the algorithms are more complex and lazy race values are sooooo easy. (How about removing all images??? Just backgrounds and social media posts. Too hard, I know. Black and white is so easy for the lazy man.)

    Science died years ago. Maybe you should refer to the current practice as the post-science era.

    Am I racist for hating Kenye West destroying a ranch with a 52,000 sq ft mansion and two underground garages, trashing a cute old west town and making Wyoming famous via latching onto a billionaire? I hate Kim too, but she’ll divorce him within 5 years, so it’s not such a big deal with her. Actually, if he was white, green, purple or orange, I’d still hate him. (I totally disdain the white drooling idiots in Wyoming that are trying to look important by having a crap billionaire living in their state. Guess I’m not racist.)

    (Minuteman: So the positivist never questions the interpretation and choice of data? Interesting. Sounds exactly like all of the sciences. I see no difference.)

  4. There’s a lot of non-entities in that long, long signatory list: “undergraduate”, “PhD candidate”, random people of all walks of life. Not that there’s anything wrong with these stations, but it looks like they basically dragged people off the street in order to make it look more impressive. Compare it to yesterday’s infamous open letter to Harper’s about free speech, which is signed by genuinely accomplished (if often awful) people.

  5. Democrats caused the Blacks to be in the state they are now. Two quotes from LBJ regard the passing of the Civil Rights Act support this contention:

    —“These Negroes, they’re getting pretty uppity these days and that’s a problem for us since they’ve got something now they never had before, the political pull to back up their uppityness. Now we’ve got to do something about this, we’ve got to give them a little something, just enough to quiet them down, not enough to make a difference. For if we don’t move at all, then their allies will line up against us and there’ll be no way of stopping them, we’ll lose the filibuster and there’ll be no way of putting a brake on all sorts of wild legislation. It’ll be Reconstruction all over again.”

    —“I’ll have those niggers voting Democratic for 200 years”

  6. Scrolling through the list of signatories….my goodness that’s a long list. All the usual suspects of our betters.
    I wonder what percentage actually read the letter? My AI algorithm predicts “very few”. Reading or critical thinking is no longer practiced, they simply know to obey and be on the “right side of history” (my personal most hated phrase of the New Age).

  7. I have to say first that the paper they’re complaining about is ideal from their point of view: it does seem sketchy. Facial characteristics predict criminality? Very much like phrenology. But with current “AI” methods (and I recently worked extensively in AI) it’s very difficult to distinguish correlation from causation. And also difficult to even know _what_ the algorithms are keying on. It’s been discovered multiple times that a “model” keyed on background characteristics, not the foreground – e.g., if the “criminal” pictures are all police mugshots while the “non criminal” picture are normal selfies – the model will latch on to features you’re not even thinking of … the color of the wall the person is standing in front of, etc. etc.

    Anyway, the left is desperate to completely close off entire avenues of inquiry as beyond the pale, not even scientific to look at it at all. For years the fight they were fighting is genetics vs. intelligence (or, genetics vs. success in life for some any reasonable definition of success). Same reason: The horror of discovering the science of genetics means that obvious measurable racial differences are not just limited to athletic skill, but extend to other areas …

  8. Sheri – blacks commit crimes, especially violent crimes, at a rate that is ten times that of white men. They commit more crimes both per capita and in raw numbers. One out of three black men is a convicted felon.

    It’s because they’re black, with everything that goes with that, both nature and nurture. Blacks are violent and dumb. They always have been. They always will be. It is their nature. Race is real.

    Heed the wisdom of our grandfathers: “The Negro is at your feet or at your throat.”

  9. This is good for us, as the capacity for using this kind of technology erodes their ability to use it against us erodes too. The technology can also be reinvented fairly quickly after we win, the core research is widespread and well established. We will just have to retrace the last few years of progress, if even that.

  10. An algorithm that rated intoxicated drivers as higher risk for accidents would be “racist” against drunk drivers. Sure, blood alcohol level (BAL) might be the single most correlative factor, but you could create more complex algorithms that did not rely on BAL. Using BAL is just lazy.

  11. I can’t tell from the linked article, but I’m going to assume that black and white weren’t the only races thrown into the data set. If so, relative crime probabilities would exist for all the races versus each other. If you really wanted to set the current Left’s teeth on edge, you’d just focus on the black vs white ratio. It would be interesting to see the all the data, how it was analyzed, and what issues were included/ignored. Certainly that information would allow a more balanced discussion around the whole concept.

    Even that exercise is removed by eliminating the ability to ask certain questions.

  12. Race is a lazy, political way of dealing (apparently a lazy statistical choice also)

    No at all. If one group is 10x more likely to be X then whatever defines that group is an important factor. Ignoring it makes the computed probability less reliable.

    Facial recognition works best on white faces largely because of training with predominately white faces (and other less important reasons such as intra-image contrast). The only way to combat this “bias” is to train with an unrealistic database (say, even distributions of back and white) making the resulting model less reliable.

  13. The whole purpose of AI is to discriminate based on the features of the individuals so as to achieve accurate prediction. It has no other motivation and will only reveal how humans already act, including any potential racial bias in terms of incarcination. Of course, if the predictions of AI are acted upon, this will perpetuate the bias.

  14. A quick search online reveals that Springer rejected the paper for reasons independent of the complaint discussed in the blog.

    As the paper was about using facial recognition software to predict if face-imaged individuals would become criminals. That is pseudoscience – any AI program applying any kind of AI algorithms would be identifying correlations about facial features and behavior, and from that extrapolating to forecasted outcomes. An analytically sophisticated analog to predicting presidential elections based on the height of the candidates.

    Correlation does not mean a bona-fide cause-effect relationship exists. Yet that is what that rejected paper appears to have had as its fatal flaw.

    Which makes this particular blog entry interesting:

    Briggs has rightly wailed about the error of confusing correlation with causation many times. Why not point that out again here?

    In this case the 2000+ social justice warriors critiquing the paper were correct about the underlying [pseudo]science being debunked – why not give that bit of credit where due, AND rebut their whiny activist misplaced socialist propaganda presented as objective rebuttal?

    Because that’s how polarized we’ve become.

    It’s all-in on a side, all the time. No middle ground.

    Peer review worked and rejected that paper (egads .. “peer review worked”!!).

    Briggs commentary, however, to rebut the SJW socialist propaganda, effectively endorsed the paper’s flawed premise by overlooking that bit of common ground entirely. So it appears have most reading & commenting (David-2 being an outlier in showing some objectivity).

    When most of us are putting political/social values ahead of objective review of a paper — sides taken and sides opposed trumping real analysis of the facts — we are all in trouble … hardly anybody is really thinking or listening.

  15. I knew we had to have a vile, contemptible racist or ten here. Thanks for showing up McChuck and reducing blacks to nonhuman status like your forefathers probably did. I have zero respect for you (and I know you don’t care—racists gotta hate and they don’t care if they are evil or not). I can see why the nation went to hell with all the evil horrible racists out there. I am convinced whites have this coming after reading things like you write. YOU are the animals.

    DAV—Using ONE major factor detracts from the rest. Michael Mann thanks you though for teaching people things reduce to one factor, like CO2.

    The internet has convinced me that it may be true Trump followers are the most racist out there. The white, rich, elite call blacks animals and think it’s fine. Then complain that blacks call whites animals. Hypocrites. Humans are just horrible.

  16. @Ken

    The entire field of Machine Learning is pseudoscience but… it works. When Good Old-Fashioned Artificial Intelligence gave way to new Push Data Through Neural Networks And See What Comes Out we left science for sorcery. Nobody knows what kinds of correlation AI makes, nor why but it works. We can only see conclusions. It does not have to make sense to us, it just has to prove itself in the field.

  17. @Minuteman – If only you knew! (Maybe you do…) I worked at an AI startup (which was purchased/absorbed by a major company) and if you saw how the sausage was made … what “engineering” process went in to developing models … the “ML scientists” called it “running experiments” as if it was a laboratory science and they just had to dial in the temperature of the condensers to get maximum yield but to me it was indistinguishable from trial-and-error … in fact, it was more like throw a bunch of random parameters against the wall and see what sticks …

  18. And yet… Charles Darwin’s nonsense theory of evolution will still be taught in schools and used as the basis for science and to cancel scientists who so much as express criticism, skepticism or rejection of the ‘accepted science.’

    Establishment Scientific stupidity didn’t just start with today’s lefties. Yesterday’s lefties have been dismissing fact-based evidence and substituting fantasies since the time of Galileo to dismiss the Catholic Church and the ancient cosmologies that all inferred Divine origin for ‘order out of hagan lio’ naturalist BS.

  19. HU facial recognition software predicts criminality……throw it out!!!

    This software may well find similarities in facial features to those who have already committed crimes, but in no way does this software allow for the workings of God’s Holy Spirit within the human spirit.

    Why should people even try to live honestly if software is their prosecutor, judge, and jury???

    No, this is not a good usage of science.

    And to those who are black of skin, I am so sorry for the unjust and cruel comments above. Those words do not speak the truth. And Jesus said that it is what comes out of the mouth:

    “16 And Jesus said, Are ye also yet without understanding? 17 Do not ye yet understand, that whatsoever entereth in at the mouth goeth into the belly, and is cast out into the draught? 18 But those things which proceed out of the mouth come forth from the heart; and they defile the man. 19 For out of the heart proceed evil thoughts, murders, adulteries, fornications, thefts, false witness, blasphemies: 20 These are the things which defile a man: but to eat with unwashen hands defileth not a man.”
    Matthew 15: 16-20.

    God bless, C-Marie

  20. 400 years ago ,we forcibly brought blacks over from Africa into slavery. That ended in 1865. For the past 155 years, blacks and their increasingly less black descendants ( because of mixed race copulation) have struggled to become part of the white man’s world. Some have succeeded, most haven’t. It was made much more difficult because of systemic racism for the period from 1865 to 1985 ( 120 years). Since then systemic racism has mostly disappeared (statistics have proved this). I say mostly because there are still 2 examples of it (less a proportion of black police officers to non blacks and longer prison sentences for blacks than non blacks. However even these instances of systemic racism are decreasing every year. However this does not address the real reason why black men commit much more crime on a proportion basis than non blacks. The real reason is the gang culture of non whites. You can see it in the music, the aspirations of most 12 year old blacks, the breakdown of families in the black community, and the number of single parent families. Unfortunately the same scenarios play out in indigenous communities all over the world. The Democrats solution to ending this problem is to enforce Communism on all of us. That solution would unleash the hell of modern day Venezuela on our societies. I do not pretend that I have a solution to break the gang culture of non whites, but to ignore that it is the problem is to ignore history.

  21. Sheri,
    Using ONE major factor detracts from the rest

    Is cabin fever taking its toll?

    Who said anything about ONE factor? I said “an important factor” without stating what others might be as well.

  22. DAV

    I think COV is messing with us all.

    A reason why I really didn’t want to comment on this post. Tried to be funny .. didn’t work.

    If I would’ve commented my initial thoughts, I might have merely pointed out that it (the “AI” work) reminded me of old psychology textbooks discussing cranial sizes and body types and IQ and rot like that.

    Yeah! I think our host was TRYING to show the hypocrisy of the “ENEMY” embracing “science” until “science” offended them. Unfortunately, our host now SEEMS to be endorsing a “science” HE rails against, which Ken pointed out so aptly. (I guess the “enemy of my enemy”.) Unfortunately, glass houses and stones seem also to apply.

    It’s been a long twenty some weeks

  23. john b(),
    Ken is doing what Ken does best: missing the point.

    Briggs doesn’t rail against AI. Instead he points out that AI is a misnomer (that is, not intelligence) and is nothing more than yet another statistical model. Why would he rail against statistical models?

    Ken,
    the paper was about using facial recognition software to predict if face-imaged individuals would become criminals. That is pseudoscience – any AI program applying any kind of AI algorithms would be identifying correlations about facial features and behavior

    Yes. That’s what neural networks do. All of them. They don’t identify causes. They merely predict. The thing about neural nets is that the designers really do attempt to verify the predictions.

    Correlation does not mean a bona-fide cause-effect relationship exists.
    So what? You can predict the day’s temperature will shortly increase upon seeing the sun rise. You don’t need to know why. It just has to be reliable.

    Oddly, facial recognition works better on white faces. Networks that “fill in the blanks” — that is guess at the blurry parts of a poor resolution photo — tend toward inserting white features. Why? (1) because they were trained on predominately white faces and (2) in the U.S. at least, it’s the safer bet as blacks only make up about 15% of the population.

    NNs are really good at finding easy correlations. If they are identifying more blacks than whites as future criminals then they aren’t telling us anything new. What’s more interesting is that they can differentiate between blacks and whites at all.

    Springer should publish the paper. It should stand or fall on its own merits and not because a bunch “concerned” idiots scream “Racist!”. If it doesn’t work (meaning it is unreliable) then it will get the drubbing it is due. The problem apparently is that it likely is a good predictor so the response is what all good Lefties do when faced with things that challenge their world view: yell “Racist!” and plug their ears.

  24. “data scientists, and sociologists”

    What the infernal fornication are these two groups doing in the same room… Sociologists know less than nothing about anything…

  25. For years I’ve told people, if you want to reduce gun crime, don’t allow blacks to have guns. Instant over 50% reduction in gun crime. RASCIST !!! (same as AI)

  26. Did anyone here yet think of what makes someone a criminal in the US? And about culture?

    I think it’s possible that most laws that are broken, which were created by a predominantly white country, may be broken because people who aren’t white have a different cultural view of what is acceptable and what they need to do to survive. It’s possible that non-whites feel they need to go to more extremes to survive and the US laws are based on what white culture deems acceptable.

    Therefor, maybe being a criminal should not be considered “bad” but something that just is what it is in relation to the laws of the country that is being discussed.

    So maybe when you think about the statistics that blacks commit more crimes in the US, you can think about how their family legacies and tendencies are rooted in different cultures than most white people and decide that it doesn’t make them bad or lesser of a people, but just not ones that fit in with the laws of the US.

    In conclusion, I think all should take good and bad out of their perspectives, else they are being an almighty judge of others who are also human equals. Good and bad is a matter of opinion and evaluation of the circumstances.

  27. It is not culture, it is purposely fatherless homes which are at the root of the problems, for black and all colors of children. And more black children grow up in such conditions than children of other colors including us so-called white-skinned people of whom there actually are very few who are truly white….true Nordic peoples have the white skin…..most of us have colored pigmentation. We “white skinned” people ought to be named varied-colored people, for not many of us have only Nordic white.

    There was a time when races were not categorized on skin-color, but they were termed by country, as the Germanic race, Italian race, English race, etc. . And there was a time when Europeans had not yet much intermixed with peoples of color. But now, we are almost all a mx of one sort or another.

    Theft, stealing, robbing, looting, murder, beating up, intentional slander and defamation, invasion of person, drug use leading to the aforementioned and more, are not acceptable anywhere the Ten Commandments are honored whether the Commandments are known as such or not.

    God bless, C-Marie

  28. “People in glass houses should pull the blinds before they change their trousers”
    Sike Milligan. Calendar at the weekend in my Mum’s kitchen.

    AI isn’t as crazy as it first seemed to me

    When “we” “I” “one” notice patterns in complex situations that then repeat, it leads to a sense of anticipation of the outcome for reasons not necessarily understood. This is what can feel like intuition, hunch or even something more objective, I.e more memory based. I think. The brain imprints a memory and the sensitivity is increased so the memory is triggered. That can be a well formed memory or just a simple pattern of association

    I noticed this during hours and hours of assessment and diagnosis to the point of beyond routine. Patterns (which help investigation, saving time)appear for reasons not necessarily officially understood but they are useful .

    It’s the same as all people do when they have a hunch but it’s more than a hunch, you just don’t understand the why or the cause of the association which tends to repeat itself. That is what is called experience. That was always the part which interested me but there’s never enough time to follow it up. Now, with computers, this kind of information can be fed into a computer which is accessed by thousands.

    AI is definitely on to something because it’s doing the same kind of thing as the brain. Humans can’t be fully objective about themselves so it’s also hard to understand a system that is mirroring some of the same functions. Perhaps it does find obvious, already known traits or groupings regarding the black white issue bu when it comes to more complex, and numerous types pf information I can see it is of some use potentially. Even in discovering a pattern not yet noticed.

    The AI isn’t the same as a brain doesn’t matter. It’s still another kind of biomimicry of thought

  29. Bnon relativizes thus:
    “I think it’s possible that most laws that are broken, which were created by a predominantly white country, may be broken because people who aren’t white have a different cultural view of what is acceptable and what they need to do to survive. It’s possible that non-whites feel they need to go to more extremes to survive and the US laws are based on what white culture deems acceptable.”

    And if it’s true that non-whites feel they need to “go to more extremes to survive” [i.e. commit crimes] in a country founded and developed by European men, what does that tell you about immigration and multiculturalism?

    But it’s not only the US: people of African descent commit a vastly disproportionate amount of violent and property crime in the UK, Canada, Australia, Sweden, and a host of other countries. It’s true across time as well: even when most African-Americans grew up in intact households, they committed crime at higher rates than other races.

    Rather than fretting about “racism”, whatever that is, perhaps we should consider that African-descended crime is such a widespread, robust, and repeatedly demonstrated phenomenon that it’s basically impossible to solve, and that the best solution is separation.

  30. @Murray

    Yes. I agree. That’s one thing about the boiling pot of the U.S. it that we’re not mixing. We’re being separate and only conforming to our own family’s ideals. Why else are there places like Little Saigon next to where I live. Because Vietnamese and other Asians only prefer their own culture, but want the benefits that the U.S. rules and laws allow. It’s like having multiple countries existing within one. I think we should be separate, too. Otherwise the only way to solve it is for one culture to dominate the other, because everyone’s scale of morality is relative to what they have known and experience. Trying to coexist seems to just spark many scuffles and arguments because each side would rather fight for their own beliefs rather than conform to another’s. And why should either of them? So separate is what I see as the best option, too. But how can it be done now that all of these cultures are mixed?

  31. When I said that one third of all black men were convicted felons, that was a fib. It’s really more than that. One third of black men are in prison or on parole for felonies. Please note that 95% of these are cases that were plead down from more serious charges.

    Peter said – “racial bias in terms of incarcination”

    Incarceration is based on criminal activity. Blacks are jailed less for their crimes than whites are. Blacks commit vastly more crimes per capita in the US, and more in absolute terms.

    Race Population Crime
    White 5/8 1/8
    Latino 2/8 2/8
    Black 1/8 5/8

    Sheri, et al. can’t handle the truth. None so blind as those who just won’t see.

  32. I saw a blurb on Instapundit that linked to an article about how face masks are breaking facial recognition algorithms. I’m not a fan of either masks, or facial recognition technology, but it’s interesting to note that the “left” has generally been against such technology but is now the group pushing mask wearing on everyone. Weird, huh?

Leave a Reply

Your email address will not be published. Required fields are marked *